Hello XMG I live in the UK and am looking at purchasing this laptop:) there seems to be some confusion over where to buy this is you live in the UK! Please can you help? Thankyou
Same! I have the same config and got mine delievered last night. Did it through XoticPC and they were able to repaste the CPU and do a few adjustments, but I'm loving it!
Maybe it would be an interesting idea if you could make another video with the latest BIOS ( 1.07.04RTR3), as well with some instructions on ho to make the update. As far as I know the last BIOS updated the performance profiles.
Already working on :) I got a lot to do for my job till end of the week, but I can imagine to publish a "first month" review video next week. I want to investigate the new BIOS, new PPT values, overclocking, undervolting, fan noise and some more memory tuning (3200CL16 vs 3200CL19, 3200CL22, 3000, 2666) and possible GPU undervolting to match the RTX 2070 Super's performance. A lot of benchmark runs needed for that. Maybe it takes some more time.
i would not expect more than 90 minutes in best case. Under some loads maybe even just 60 minutes. Not tested though. The equation is simple, you got a 62Wh battery attached to the Apex 15. The power consumption by the RTX 2070 if it is idling (windows, websurfing) is 15W, the CPU sitting in 88W performance mode reports to use 28W while writing this text. Plus all the peripherals like display, USB power, WiFi, maybe another 3-5W. In total we have 15+28+5W = 48W while web surfing. That means the battery would last 62Wh/48W = 1,29h = 77 minutes. You can also push down the CPU's maximum power (PPT) down to 28W, which reports up to 22W while websurfing. So 6W less, or 62Wh/42W = 1.48h = 88 minutes Maybe there will be more power saving features in future. But that is what you can expect right now. As you said, desktop replacement with desktop performance. Not an ultrabook ;) It's a trade-off
@@TechModLab Wow I find it both surprising and disappointing that the CPU scales down so badly. I mean... never going below 20W or so, even when choosing a very low power profile...
I'm almost tempted to get my old watercooled 2600K out of storage to see how much it uses during idle... Update: Ah, why bother when anandtech did it many years ago ;) www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/21 Kind of looks like AMD has never resolved that issue over all these years?
@@notthere83 True, in idle Matisse (Zen2) is worse than Intel, but better under load. Kind of a trade-off. The thing is, the power needed by the CPU cores can be pushed down to 3-5W, but the SOC power is drawing 8.5-10W at all time. In idle, in load, always. That sums up to 12-15W. I don't know where the other 10W are getting lost, though. But AMD's Renoir mobile Ryzen 4000 are much better in idle. Wonder what's different. Could be that Renoir is made in 7nm as one chip instead of combining 7nm core chiplets with a 12nm IO die, like Matisse Ryzen 3000 is. I wonder if Zen3 will change that. But even with a more energy saving CPU in idle, you still got the 15W of the idling GPU. You can only mitigate that if the CPU would be an APU containing a smaller GPU which could be used for everything but gaming. Just like Nvidia Optimus does.
Just got a thought looking at the video and the loud spinning fans, as I also saw some videos on water cooling laptops and the heat pipe arrangement of the Apex 15, how about this: You can see the heat pipes on the Apex 15 are directly under the lower cover (1:15 to 3:50 in the video). For hardware hackers It might be an option to just cut the plastic housing open there and than create a kind of upstand for the laptop which has a large watercooled plate at the same spot. Than just pose the laptop on that cooled plate. Copper to copper. One might not even need thermal paste as the area is very large. It might even work to some extend to just put the laptop without plastic base cut on such a plate, still I do wonder about condensation issues than. Besides cutting the plastic base cover open that would not make other mods to the Apex 15 necessary. You could even add a cover to close the hole when using the Apex 15 on the go. Thoughts & comments are welcome. A suggestion to XMG, perhaps it might be an idea to create something like this as a sort of docking station. Ship it together with a modified bottom cover than.
I know the answer for this question requires some testing but how are the temperatures and fan speeds CPU running 65W vs 88W? How much lifting the back of laptop affect to the temperatures? Oh and great job with the video. I really like your structure how you constantly inform what is going on now and what is coming next. Some bigger UA-cam "stars" hide the relevant information from their videos to keep people watching the video for a longer time (=means more profit to them) but thanks how you did it it was just pure enjoyment to sit and watch it. Maybe add the timetable to description?
Actually, it seems like UA-cam has introduced a feature where you can set time points right in the video - Hardware Unboxed is doing this for instance. I would recommend doing that, so you don't have to render the video with time stamps and then also put them into the description or something.
Add a timetable? Done. :) I couldn't find a significant difference in CPU and GPU temperatures when the laptop's back get's lifted up. But the RAM temperatures decreased by 1-2°C in high load situations. The SSDs could benefit, too. Not tested. I was investigating the temperature difference in 88W PPT and 65W PPT for the video, too, but didn't notice any difference in temperatures using the "auto" fan setting. XMG state to use a temperature target for their fan control, so I think that's the reason. SOTTR is using the CPU and GPU in a balanced way, so the PPT actually used during the SOTTR benchmark is around 55W. I guess that's another reason why 65W/59W PPT and 88W PPT makes no real difference. Just in some spike load situations. Check out the bottom of description of the video. I made a mistake and tested SOTTR in entertainment mode with slightly different settings. Performance in SOTTR is nearly the same in performance and entertainment mode. On the other hand the difference in CB20 was big. CB20 with 59W PPT was noticably quieter. In the 32W and 28W the fans were not spinning up at all. Kept sitting at their idle speeds.
@@Juha.Latva-Somppi I said "I couldn't find a significant difference in CPU and GPU temperatures". That means the temp difference was less than 1°C. The difference could increase over time though, when the table underneath heats up and heat up the incoming air, too.
Maybe. The M.2 slot for the PCIe 3.0 SSD is directly connected to the CPU. I wouldn't really bother with *doing* it, but I'd sure like to see someone else try. We could even lend out a GPU if it's in the EU. // Tom
Erstmal, vielen Dank für diesen ausführlichen Review! Wie sieht es aus mit der Akkulaufzeit? Wie viele Stunden könnte man mit nur Web browsen, Code Editing, usw. erwarten?
Gerne! Ich habe die Frage so ähnlich schon dem Nutzer Shield Storm beantwortet. Dort habe ich von maximal 90 Minuten und bei leichter Last 60 Minuten geschrieben. Ich habe das alles aber nicht getestet, nur anhand der gemeldeten Leistung der Sensoren abgeschätzt. Ich war da anscheinend aber noch zu optimistisch. Die Akkulaufzeit wurde jetzt von Notebookcheck mit maximal 73 Minuten im Idle und 62 Minuten Websurfing bestimmt. Das ist wenig und ist folglich nur was für Leute mit Stromstecker in der Nähe. Die Zen 2 Desktop CPUs sind im Idle leider nicht stromsparend.
Noch ein Update dazu. Es könnte sich um Unstimmigkeiten mit dem Notebookcheck vorliegenden Notebook handeln. Sie planen einen Folgetest. Notebookcheck sagt, dass sie 60W im Idle haben, das kann ich bei meinem Apex 15 so nicht bestätigen. Bei mir sind es tatsächlich 45-48W. Könnte sein, dass die verschiedenen CPU und GPU Optionen auch unterschiedlich hohe Verbräuche haben.
Unfortunately no. But I expect some benchmarks coming soon, f.e. on notebookcheck regarding those Refresh GPUs. I bet even inside a XMG test notebook. There are no reviews out yet because there was a NDA about the new Super and Refresh GPUs till mid April. I think this is the reason der8auer wasn't talking about the GPU inside his prototype sample at all.
hello, does you display works with nvidia drivers 531.18? I get my screen backlight off with drivers above 522.30 and newer. The screen just turns off after driver installation on apex 15, the problem is only in backlight of lcd, it even has image, what can be seen with an external light source.
Excellent review. Is the GPU soldered ? Thinking about getting one with. 2060 refresh then upgrading it to a 2070 if I ever need to. Really like that they have made this chassis tinkerer friendly.
@@mudanenadaara MXM is a loosely defined standard for modular/removable GPU boards in laptops and a few small form-factor PCs. We are using it in XMG ULTRA series. ZOTAC has used it in MAGNUS EN1080K. It's not well supported by NVIDIA (and virtually not supported at all by AMD these days) and it often does not survive upgrade attempts across multiple generations due to VBIOS and heatpipe placement incompatibility. And yes, it adds a lot of R&D and validation cost while being less efficient than an integrated solution. The lack of efficiency comes from the interface of routing everything through a physical connection, but also from the limitation this puts on the general GPU board layout. // Tom
Hi! Thank you for the in-depth teardown and review! May I ask if using this laptop purely for gaming, will there be a noticeable difference if I were to install games on the faster M2 top slot rather than on the slower M2 bottom slot?
depends on the load, but for daily tasks you will not notice it. The real benefit for gamig is minor in my opinion. But using the faster slot could help to prevent micro-stuttering.
This is boxed, wrapped and also shares the same chasis as the Acer Predator Helios. Has to be in with Acer or getting the same manufacturer to construct them. The cooling pipes look much better than the Acer Predator series. You can get an Alienware Area 51m R2 with a better screen and similar specs for about a grand cheaper than these. Great to see someone breaking in to the market, but they have a long way to go on design. Hard to charge 4k for a laptop that looks like it's from 2010.
New video is coming next week, will address that, too :) In short, I still like it a lot. I spent a lot of time tuning the components, not because it is necessary, just because I like to do that. After tuning I can pretty confidently say that I was able to improve the user experience, by lowering nosie and temperature levels and keep the performance, maybe even improve it a little. Stay tuned. I chose the 3700X because the 3900 was 150€ more expensive the day I ordered. There was a pretty noticeable price drop on the 3900 after two weeks. But still I am not unhappy. The 3700X seems to be a very good allrounder. I noticed in some reviews that the 3900 might run a little hotter and consumes more power as the 3700X in general. But I got no 3900 to verify that, impossible to clear that up. On the other hand I plan to upgrade to Ryzen 4000 'Vermeer' Desktop next year anyway, so I was looking for the best bang for the buck.
@@TechModLab thanks, very clear. After your impressions about the 3900 (despite you didn't try it) I understand that you don't recommend the 3950x at all,right? Maybe at this point if someone is looking for a very powerful laptop similar to a desktop the XMG Ultra could be is best choice ?
@@marcolondero Didn't say that. The 3950X can be a very different story. In fact the 3900 and 3900X is kind of a "failed" 3950X and is recycled. It could be - not sure just thinking here - that the "silicon quality" of the 3950X is better and it runs colder at the same performance level as the 3900, simply because it got the more efficient and more quality CCDs. In case you wonder, CCDs are the individual Chiplets AMD uses on their Ryzen CPUs. 3600 and 3700X got only one CCD, 3900 and 3950X got two. Threadripper uses up to 8 CCDs in their 3990X. Each CCD is tested after production and gets a quality label. Lower quality CCDs end up in the 3100, 3600 processors. Highest quality ones in 3950X. The other CPUs are in between. Lower quality = less max clock capable of maintaining stable, or uses more power than others for same performance. The only thing which hold the Apex 15 back from being a "very powerful laptop similar to a desktop" as you say, is the RTX 2070. The CPU is superior to the Intel ones, not in gaming, okay, but in general they are. The Apex 15 is more lightweight and more efficient than the latest Ultra series. The new Intel 10th gen processors get hot AF. And the power supply of the Ultra series is way bulkier than the one used by the Apex 15. Hard to compare both. The Apex 15 formfactor and weight is more a real laptop than a desktop replacement as the Ultra series. But I think the Apex 15 can easily keep up with the latest XMG Ultra 15 with a RTX 2070 inside.
Hi, thanks a lot for the review! I am usually working on higher performance (3D modelling and rendering) and mostly there are also other people in the office. Do you think the noise levels of Apex 15 are okay, or could they be disturbing for those not wearing headphones? Is it much different from other laptops? Thx!
Depends on the load of course. Hard to say to be honest. The notebook itself can get very loud if you max out the components, f.e. CAD production render (more than 51dbA possibly, i got no noise meter). But you can work in lower power modes in office (using ControlCenter or the Hotkey Fn+3). As shown in the video there are some clear sweet spots between performance and power dissipation (16:30). XMG stated in the german forum that they are planning to lower the entertainment mode's PPT down to 45W. I expect very good results around 4000 CB20 points from that mode while maintaining the notebook quiet. Let's wait and see how it works out. :)
It gets REALLY loud if needed to cool itself use carbonaut/cryonaut on cpu and GPU as that keeps my temps low, my laptop idles in the mid to upper 40s with a pretty much silent fan curve until both cpu/gpu get in the 70s
@@sbeezynukka thanks for sharing. so you're saying it's pretty quiet up till 70 Celcius? Would you be able to measure noise levels (e.g. using a noise meter app?) And let us know? Particularly keen to know how "loud" the quiet mode is and what performance you get at that level.
Nice review. Couple of tips: Try locking your cpu at a fixed speed at lowest voltage possible. From testing with a desktop 3600 I found that the following is 100% stable: 3.6GHz@0.97V, 3.8@1.04, 4.0@1.13 which made idletemps much lower and more stable, powerconsumption dropped about 20% at 4GHz, 30% at 3.8GHz and 40% at 3.6GHz. As for ramtuning I bet you can run your ram at 3200cl18, try lowering trfc to 280-330 if it's B-die, that will improve gamingperformance a lot :)
Sounds interesting! You mean fix all cores to the same frequency in manual mode? There is a chance that it will result in less performance than just lowering the PPT. There was an interesting investigation made by pcGamesHardware, a german website. Do a quick websearch for "Ryzen-3000-Package-Power-Tracking-statt-Undervolting-1295054", you should see their website as first result. Yes, I think I could run my RAM at 3200cl18, but we are limited to the default 1.2V SO-DIMM voltage at the moment. There is no DRAM voltage setting in BIOS yet. I maxed out the Corsair Vengeance 2666cl18 (18-19-19-19-39-58) at 3000MHz (1500MHz mclk), but it went unstable in Prime and memtest after 1-2 minutes. It was fine for gaming and making this video though. I got my hands on some faster 2x16GB G.Skill Ripjaws 3200cl18 rated for (18-18-18-18-43-61) at 1.2V using XMP, which is great, much better than the Ballistix 3200cl16 der8auer used, they are rated for 1.35V. Unfortunately my G.Skill 3200cl18 are not stable above 3000MHz. 3000MHzcl18 (18-18-18-18-39-58) is running like a charm. They are based on Hynix C-dies, so I am stuck on some quite slow Trfc sitting at 560. 525 was unstable unfortunately. I reached the point of maxing out the 1.2V DRAM voltage. Let's see what is possible with a tuning BIOS and more DRAM voltage. And I will keep an eye on the upcoming Samsung 3200cl19 1.2V SO-DIMM. They could provide better timings, or worse. Relax and wait. Cheers
@@TechModLab Yes. You will lose a bit of single core performance, but in general perf will be similar since all core run close to the same and thermals/noise is much better :) One guy I talked to ran his 3950X@0.95V which in his case improved performance and lowered thermals :) For the 3700X I would test and find a frequency/voltage at which you are satisfied with perf and nouse/thermals :) Also undervolting the GPU with afterburner and frequency-curve you can improve thermals and noise quite a bit at same performance.
@@TechModLab I have tested a lot of PPT-limit vs fixed core vs undervolt. Undervolt is barely better than stock, not worth ir, PPT is very good if you like high single score, fixed freq/voltage is good if you can sacrifice a bit ig single core for better multicore. Thermals/consumption is best at fixed freq/voltage at 3.8GHz or lower, PPT is very good around 50-70W, below 50W the frequency drpps hard. The entertainmentmode is very good in that respect, your graphs show that too :) I tested 3.8GHz@1.04 vs 65W PPT in Cinebench15: 65W PPT: 193S/1480M 54C 67W avg 3750MHz avg. 3.8@1.04V: 183S/1510M 60W avg, 51C. On my setup single score drop by 5%, multicore increase by 2%, consumption drops by 11% and 3C tempdrop. Depending on your scenario, PPT or fixed freq is best, stock or UV is not worth it.
Nice vid. Question: why did you choose the 2666 Mhz DDR4 RAM instead of the Crucial 3200 Mhz? Does it really improve performance having a lower CAS latency?
Because I didn't like any of the available options when I ordered. So I chose the cheapest. I planned to order my own faster 32GB kit some days later. I didn't like them because the max available CAS latency was 22 for the 3200MHz one (this 22 is the CAS number for example. Just one of the important ones. There are many.) Yes, those timings are important and can have a great effect on speed, because it determines how fast the CPU can save and aquire data from DRAM. It is not easy to give a rule of thumb here, because there are more effects to take into account than just raw memory timings. But let's say 3200Cl22 is half as fast as 3200Cl11 (Cl = CAS latency), if all the other timings would be halfed, too. So "real" 3200Cl18 memory could be up to 23% faster than 3200Cl22. That does NOT mean that your cinebench scores or in-game FPS will be 23% higher! It strongly depends on how much data is transfered between CPU and DRAM. The more data needs to be transferred the greater the impact. Cinebench for example got a quite low impact. The other factor which is really beneficial for speed is to use an higher DRAM frequency, because of the infinity fabric clock. It is the clock that is responsible for the communication between the IO die and cpu core chiplets inside the Ryzen 3000 processor. The higher the DRAM frequency the faster the CPU's connection to "the outside world". To make things short, in SOTTR high settings you can expect 1-4% less performance when using 3200CL22 instead of 3200cl18, maybe more on minimum FPS which are a big deal for possible micro stuttering. NOT TESTED though! Maybe I will make a followup video about that. That answer was long... Hope that was not to much information. :) You can also do a quick search on youtube about "ryzen 3000 ram speed". I can recommend GamersNexus and linustechtips' videos about that topic. Cheers
@@TechModLab I'm looking for some faster ram on some website for this laptop too but all I got are 3200MHz with cl20 or cl22 , may I know what ram you are going to use?
I did quite a bit of research about this myself and found this really extensive polish benchmark that shows how much more important frequency is compared to latency: www.purepc.pl/pamieci_ram/test_pamieci_do_procesorow_amd_ryzen_taktowania_vs_opoznienia?page=0,16
@@WenYSiow I am using G.Skill SO-DIMMs now, 2x16GB ones (G.Skill F4-3200C18S-16GRS). Unfortunately they seem to be out of stock at the moment. But there are still 8GB sticks available, if you think 16GB total is enough. There are also 3200CL16 8GB sticks. But my 16GB sticks originally running advertised with 3200CL18 are capable of running CL16, too. There is good overclocking headroom if you get lucky and get some well binned memory dies.
Thanks for the video, it was very interesting. I've just received my Apex 15 and I must say that I'm loving this laptop. However this little feller is making a huge noise out of its fans. It stays on 60-65 degrees (CPU) on idle and then it reaches 85 - 90 degrees while I start playing some games (performance mode) . Is it normal? Any suggestion on what shall I do? When I pick the quiet or balance mode in the configuration tools, the fans actually start spinning a bit lower for the sake of my ears, but the temperatures are still around 85-90 degress while playing some games. Nothing to say about the performances, on the other hand: it's beast
Hi! Sorry for the late response, UA-cam does not show up all comments in the comment overview. Well you can undervolt the CPU a little, you can lower the temperatures by quite a lot ding so. Unfortunately the fan noise is not really affected by doing this, because the GPU fan is the noisy part. Unless you can not limit your GPU load it will be noisy. Sorry. A good tip is to limit your FPS to a resonable value, like 60FPS, if you're willing to do so. I can understand peeps want 144FPS to enjoy 144Hz monitor. Check out my new video. I explain CPU overclocking/undervolting there in more detail
depends on your use case. For gaming the 3700X and RTX 2070 is better. For content creators, video editing, rendering more CPU cores are important. But I would always recommend to take the biggest GPU possible, cause it is not upgradeable. The CPU is though
New video is coming next week! Preparing some content concerning new BIOS' powermodes, RAM tuning (2400MHz to 3400MHz), GPU and CPU OC with undervolting, fan noise and temperatures. Also I want to address some commonly asked questions and missinformation sourrounding the Apex 15 and Clevo NH57AF1.
Really looking forward for that. I have the same laptop and it's a toaster. While in power saving mode everything is fine, but once I switch to Performance, I get Ryzen Master temperatures around 85C on Forza Horizon 4 (Entertainment mode is even worse).
@@stefanthis1258 Do you have the same CPU, the Ryzen 3700X? Do you have an original XMG, or a different vendor? Which BIOS version? As a little check what could be wrong you can install HWInfo, run in Sensor-only mode, rescale the window till you see the 4th column "average" value. Now start your game and reset the values by clicking on the clock symbol down below. Now play at least for a couple of minutes and report the average on the following values: CPU (Tctl/Tdie), CPU Die (average), CPU Package Power, GPU Temperature, GPU Power. Maybe we can find a clue together
@@DavideStiff Hard to say, I did not get my hands on. Performance wise yes, but I am not sure about thermals. As far as I read it could be that the 3900 runs a little hotter than the 3700X. Depends on what you do with the Apex 15. Rendering a lot? 3900 or 3950X! For gaming buy the Apex 15 together with a 3600 or 3700X.
Sorry for the delay. Video will come in 1 or 2 days. Too many data and too many tests. Cutting down the content into two or three videos. Good thing is I already got most o f the data. Big respect to all those tech channels which test components all day long, it is super time consuming. Relax and wait guys, it will be done. ;)
Bought it finally after having a rather flaky experience with a Core 15 TongFang barebone (not from XMG but the same barebone)... Now I see how bad the battery life really is... I wish there are denser battery (same package, voltage, but with larger capacity) made for this chassis!
Physically not possible. Maybe +30% at best with some high quality Li-ion cells, but there is no better battery technology available yet. The main problem isn*'t the battery capacity to be honest, it's the base power consumption. Desktop components are not made for power saving tasks.
@@TechModLab Yeah I see now. After looking around at one of those 160wH massive AC output camping batteries, even that will probably only last 3-4h with the laptop sipping 50W doing light tasks. And the price on those huge battery units also climb up real quick as you go up in capacity (plus the weight and the size gets impractical too).
So this config vs. the 16 inch Macbook Pro i9 8 core 2.3GHz and Radeon 5500M 8GB. Could probably beat the MBP by at least 20% I reckon? Also would you advice using it in clamshell mode to power up a desk setup with dual mons for video editing? Im thinking of upgrading from an old desktop rig that runs i6700 and GTX970.
As far as I can tell the video editing experience is flawless. I got a NVMe though, guess SATA would be different. Got a 4770K for video editng before myself. The Apex 15 is way faster. Playing back timeline multi-layered effects on 1080p 60fps video are no problem. The Apex 15 beats the MBP with i9 by about 33% in CPU heavy stuff like Rendering, the RTX 2070 GPU is roughly 120% faster than the Radeon 5500M Pro. Clamshell means lid closed I guess? Lid closed with two external monitors works fine. You can even use up to three external monitors using a Type-C to DisplayPort adaptor (HDMI, DP mini, USB-C /w DP). And the internal monitor as a fourth one. ;) The bang for the buck is definitely better when you choose the Apex 15. Apple is a brand that always charges premium prices for their name. The components of the MacBooks are pretty average. The only question you have to ask yourself: do you want a higher quality case and a quieter laptop, then choose the MacBook. If you want desktop performance and an overall faster experience, choose the Apex 15. Hope I could help you!
Love it! After some tweaks even more. New video is coming to address that. Stay tuned. Did you receive your unit already? I wonder how the 3900 performes in idle, light load scenarios and games compared to the 3700X.
From where did you get those two screws for the battery? Were they in the same little plastic packaging in which the battery is too? Or were they already screwed in the two holes in the battery?
@@TechModLab Apparantly, the change is mostly in the Active-state power management (ASPM) but the ODM says they are still working on further fine-tuning. // Tom
Can't tell yet. Receiving more CPUs to test in the following days. Stay tuned, there will be another video about that. But I am a bit confused about your statement, there are no overheating issues? What do you mean? Kind regards :)
I don't know. But different chips, different behavior, even if the clock speeds are the same. For example, a 10 year old processor sitting at 4GHz does NOT perform the same as a modern 4GHz processor with the same core count. The IPC are different. 2060's IPC are lower than the 2070 one
@@tomik499 Notebookcheck wrote an article back in april about that topic. They had not a lot of benchmarks till then, but they talk about an overall improvement of 7% for both, the 2060 refresh and 2070 refresh compared to the older regular ones. The RTX 2070 super is another 6-8% ahead of the regular 2070 refresh.
I was looking to buy a Sager version of this with the Ryzen 5 3600, is the noise is as bad as Notebookcheck made it out to be? I understand it won't be ultra book silent but because I use this in a living room, I don't want a jet engine either. How is the fan noise in entertainment mode with some undervolting during games?
It's not as bad as Notebookcheck claims it is. They used the extreme mode for "The Witcher" to kind of proof a point. Please check out our official reply in the comment section under the article. // Tom
@@xmg_gg Wow a reply from the very XMG! In any case, their 62 dB(A) is extreme, would the entertainment value be in the range of 50 dB(A) instead, or perhaps lower on more conservative ratings? The review cited did not go into much detail at all about it, as well as the video posted by der8auer.
@@mohammadashour692 Depends. If you want to have maximum performance close to a desktop, you have to take the fan noise as is. But if you want a quieter gaming experience you could switch to a lower powermode. Depending on the game you loose 0% to 5% of FPS average in entertainment mode, but the system will stay quieter. Maybe I will make a followup video if a newer BIOS version is out.
@@TechModLab I am far from looking for a desktop class performance, I like the silent mode as you have shown but it seems a bit too close to the eco mode, I liked the suggestion to move it forward a bit, I have read XMG I believe stating that they might look into reducing the "Entertainment" profile down to 45 watts, I could be mistaken. I was looking to keep it into Entertainment and perhaps lower the clocks a little/voltage to keep it cool. I am fine even with a 10% decrease if it means it will run cooler and more quietly, that's the beauty (and curse) of having such a powerful chip in a laptop form. I have ordered my own now, but with the R5 3600, with 2933mhz ram, will see how it goes!
@@mohammadashour692 keep in mind that every notebook brand like XMG got their own fan-temperature profiles for the very same notebook. Your Sager could perform very different in concern of performance, noise and temperature.
XMG wrote in their public discord that review units of the Apex15's successor arrived yesterday. So I would wait a couple more weeks until the successor will be available. It comes with 3060 and 3070 options and Zen3 on a B550 based mainboard. OR if you don't need all the GPU power you can wait for discounts on the older RTX20 model :)
Does not matter. You can easily overclock them to their XMP profile printed on the modules via "Ryzen Master" (it get's applied in BIOS, no autostart in Windows needed). But yea, manufacturer's specs can be quite confusing about that.
@@TechModLab Well, it doesn't matter to you. And me, now that I've read about it here. But a) in their FAQ, XMG says XMP just doesn't work. Nothing about being able to apply those profiles via Ryzen Master. And so unless people already have experience with that, they won't know... And b) it's a hassle to figure out and enter all of those specifications you list in the description. Now, maybe it's not important to set all of those? Anyway, thanks for pointing this out and I hope I'll be able to made it work on my future Apex 15 as well.
@@notthere83 you got a good point there. Those numbers can be confusing. The "unoptimized" first timings I posted in the video description should work on every G.Skill Ripjaws 3200 kit by copy and paste the values. The second timings may not, they are optimized to the limit of my modules.
@@TechModLab I'll probably just use the DRAM Calculator: www.techspot.com/downloads/7164-ryzen-dram-calculator.html Worked well for techspot: www.techspot.com/review/1891-ryzen-memory-performance-scaling/
@@notthere83 True, used that for my tuning attempts too. But it's not super straight forward, as the DRAM calculator expects you to set 1.35V RAM voltage. But we only got 1.2V. works to some extend well anyway.
@@aormonde Your order hinged on an unexpected shortage in non-German keyboards which has been resolved this week. Your order should go into production early next week. Sorry about the delay! // Tom
SOTTR Benchmark gives me the following numbers... 96 FPS (Preset "Highest", 1920x1080, Antialiasing TAA). Min - Max - Average - 95%tile 100 - 173 - 130 - 101 (CPU Game) 106 - 252 - 152 - 113 (CPU Render) 73 - 159 - 98 - 82 (GPU) There is no ultra high setting, "highest" only
@@cityhunterpro I don't own Assassin's Creed Odyssey and can't run the benchmark. The witcher 3 got no official benchmark. The only possibility to PROPERLY compare my witcher 3 FPS with other reviews is when I use EXACTLY the same scene/gameplay they used. I can't, because I don't know which part of the game they tested exactly.
Performance based on some synthetic tests and real life applications, thermals, fan noise, is there a bottleneck between GPU and CPU?. Thaks for your effort in creating a fantastic series of reviews. Waiting for CPU comparison before order one 🙂. Btw is there a BIOS options to reduce gpu performance in order to increase battery life?
@@panzerdascal1 Thanks for the input :) Any wishes on real life applications? Youncan be sire about syntethic benchs, thermals and noise. Bottleneck is hard to test tbh. Strongly dependend on application.
@@panzerdascal1 Thanks for your kind words! :) And no, there is no option to decrease or limit GPU in BIOS. I would love to see a 80W or 90W switch to be honest. Would be awesome for quiet gaming sessions.
@@TechModLab too bad, would be nice to have such an option, maybe Tom from XMG can push this feature to ODM. Regarding real life application, I would be interested in of the Adobe editing apps like Photoshop, Illustrator, Lightroom. Adobe Premier for video editing or any other for this purpose. Some applications for 3d modeling like maya, 3ds Max or Blender. Also would be nice to see some code compilation benchmarks, like GCC for C/C++, or java a.s.o. Thanks again and I am looking forward to see your results.
@@panzerdascal1 added Puget Benchmark for Adobe Premiere to my tests. But no other new ones. Testing takes some days already, don't want to extend it too much ':-D So the test will contain: Cinebench R20, Puget Benchmark for Adobe Premiere, SOTTR, comparison of idle vs. videostreaming situations with noise measurements, PPT and temperatures. Processors tested are: 3600, 3700x, 3800x, 3900, 3900x, 3950x. As far as I can tell the 2 CCD processor (3900, 3900x, 3950x) use substantially more power all the time. We are talking about more than 10W more compared to 1 CCD processors (3600, 3700x, 3800x). May not sound much but it's audible in louder fan noise and higher temperatures. All in all I would say: stick with 1 CCD processors if you don't need to run production renderers/editors and need all those cores. Makes the Apex 15 much more pleasant. Anything above 3700x/3800x makes no sense for gamers inside this notebook barebone at all. You just lower your enjoyment.
Schenker website Sep 29 2020 has NO info on amd laptops like this seems only bestware integrator is making the XMG modsls ?>???? Confusing as hell. Also will they have a US power 60hz version and can their boards accept a newer rtx 3070 card in months ahead?? I want the new card in my ;aptop no rtx 2070-80 ser
bestware is their official centralized shop for all brands: Schenker, XMG and Tuxedo. There is an US AC adaptor plug available. The Apex 15 got no upgradeable GPU in MXM format. It's soldered. If the rtx 30 series will come to notebooks they will introduce a new model together with native Zen 3 support.
@@TechModLab let's say I bought the 3600cpu and I wanted to upgrade to 3950x later on, should I change my power brick as well or can use back the original brick? Thanks for answering mate 👍
New XMG Apex 15 videos are uploaded! 1) CPU Overclocking & Undervolting - reduce temperatures and increase performance ua-cam.com/video/FtLzR-i8XmA/v-deo.html I address CPU OverClocking/Undervolting and try to improve the temperatures. Some of you observed pretty high ones. 2) Memory Overclocking & Tuning - watch how to achieve 3600MHz memory clockspeed & tight memory timings ua-cam.com/video/qCkNcZ6WEQQ/v-deo.html # part 1: results & theory ua-cam.com/video/RWVm1_JCTEo/v-deo.html # part 2: memory overclocking guide Upcoming Apex 15 stuff: GPU tuning, FAQ, temperature, noise, CPU showdown (3600 up to 3950X inside the XMG Apex 15) Cheers
This is probably the best made video pertaining to apex 15, i am surprised that not many people have seen it.
XMG is here. Ask me anything! :-) // Tom
When exactly 2070 will be available again for Apex 15?
@@chrisvashck1141 New stock should arrive by the end of this week. Order NOW! // Tom
@@xmg_gg Already did :)
@XMG how much for that big mouse pad shipped to Texas? I want one!
Hello XMG I live in the UK and am looking at purchasing this laptop:) there seems to be some confusion over where to buy this is you live in the UK! Please can you help? Thankyou
I've ordered the same config from a reseller in the US! Thanks for this video.
Which one? Did you get it yet? I'm specifically wanting the 2070 refresh GPU.
HIDevolution. I went with the 3700X and 2070 combo. Still waiting on it.
Same! I have the same config and got mine delievered last night. Did it through XoticPC and they were able to repaste the CPU and do a few adjustments, but I'm loving it!
Lot of great info here, especially the PPT graphs
Maybe it would be an interesting idea if you could make another video with the latest BIOS ( 1.07.04RTR3), as well with some instructions on ho to make the update. As far as I know the last BIOS updated the performance profiles.
Already working on :) I got a lot to do for my job till end of the week, but I can imagine to publish a "first month" review video next week.
I want to investigate the new BIOS, new PPT values, overclocking, undervolting, fan noise and some more memory tuning (3200CL16 vs 3200CL19, 3200CL22, 3000, 2666) and possible GPU undervolting to match the RTX 2070 Super's performance.
A lot of benchmark runs needed for that. Maybe it takes some more time.
Nice apex Review!
I know the xmg apex 15 is a Desktopreplacer but how long is the battery live in normal use?
(Writing,surfing etc.)
i would not expect more than 90 minutes in best case. Under some loads maybe even just 60 minutes. Not tested though.
The equation is simple, you got a 62Wh battery attached to the Apex 15. The power consumption by the RTX 2070 if it is idling (windows, websurfing) is 15W, the CPU sitting in 88W performance mode reports to use 28W while writing this text. Plus all the peripherals like display, USB power, WiFi, maybe another 3-5W. In total we have 15+28+5W = 48W while web surfing. That means the battery would last 62Wh/48W = 1,29h = 77 minutes.
You can also push down the CPU's maximum power (PPT) down to 28W, which reports up to 22W while websurfing. So 6W less, or 62Wh/42W = 1.48h = 88 minutes
Maybe there will be more power saving features in future. But that is what you can expect right now.
As you said, desktop replacement with desktop performance. Not an ultrabook ;) It's a trade-off
@@TechModLab Wow I find it both surprising and disappointing that the CPU scales down so badly. I mean... never going below 20W or so, even when choosing a very low power profile...
I'm almost tempted to get my old watercooled 2600K out of storage to see how much it uses during idle...
Update: Ah, why bother when anandtech did it many years ago ;) www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/21
Kind of looks like AMD has never resolved that issue over all these years?
@@notthere83 True, in idle Matisse (Zen2) is worse than Intel, but better under load. Kind of a trade-off.
The thing is, the power needed by the CPU cores can be pushed down to 3-5W, but the SOC power is drawing 8.5-10W at all time. In idle, in load, always. That sums up to 12-15W. I don't know where the other 10W are getting lost, though.
But AMD's Renoir mobile Ryzen 4000 are much better in idle. Wonder what's different. Could be that Renoir is made in 7nm as one chip instead of combining 7nm core chiplets with a 12nm IO die, like Matisse Ryzen 3000 is. I wonder if Zen3 will change that.
But even with a more energy saving CPU in idle, you still got the 15W of the idling GPU. You can only mitigate that if the CPU would be an APU containing a smaller GPU which could be used for everything but gaming. Just like Nvidia Optimus does.
I get an 1 hr 20mim tops with a 3600x
Just got a thought looking at the video and the loud spinning fans, as I also saw some videos on water cooling laptops and the heat pipe arrangement of the Apex 15, how about this:
You can see the heat pipes on the Apex 15 are directly under the lower cover (1:15 to 3:50 in the video). For hardware hackers It might be an option to just cut the plastic housing open there and than create a kind of upstand for the laptop which has a large watercooled plate at the same spot.
Than just pose the laptop on that cooled plate. Copper to copper. One might not even need thermal paste as the area is very large. It might even work to some extend to just put the laptop without plastic base cut on such a plate, still I do wonder about condensation issues than.
Besides cutting the plastic base cover open that would not make other mods to the Apex 15 necessary. You could even add a cover to close the hole when using the Apex 15 on the go.
Thoughts & comments are welcome.
A suggestion to XMG, perhaps it might be an idea to create something like this as a sort of docking station. Ship it together with a modified bottom cover than.
This is BEAST among laptops ))
great review!!!
Thank you for this video, it was by far the most helpful one I found
I know the answer for this question requires some testing but how are the temperatures and fan speeds CPU running 65W vs 88W? How much lifting the back of laptop affect to the temperatures? Oh and great job with the video. I really like your structure how you constantly inform what is going on now and what is coming next. Some bigger UA-cam "stars" hide the relevant information from their videos to keep people watching the video for a longer time (=means more profit to them) but thanks how you did it it was just pure enjoyment to sit and watch it. Maybe add the timetable to description?
Actually, it seems like UA-cam has introduced a feature where you can set time points right in the video - Hardware Unboxed is doing this for instance.
I would recommend doing that, so you don't have to render the video with time stamps and then also put them into the description or something.
@@notthere83 Awesome, didn't know that. Will check it out. Thanks mate!
Add a timetable? Done. :)
I couldn't find a significant difference in CPU and GPU temperatures when the laptop's back get's lifted up. But the RAM temperatures decreased by 1-2°C in high load situations. The SSDs could benefit, too. Not tested.
I was investigating the temperature difference in 88W PPT and 65W PPT for the video, too, but didn't notice any difference in temperatures using the "auto" fan setting. XMG state to use a temperature target for their fan control, so I think that's the reason. SOTTR is using the CPU and GPU in a balanced way, so the PPT actually used during the SOTTR benchmark is around 55W. I guess that's another reason why 65W/59W PPT and 88W PPT makes no real difference. Just in some spike load situations. Check out the bottom of description of the video. I made a mistake and tested SOTTR in entertainment mode with slightly different settings. Performance in SOTTR is nearly the same in performance and entertainment mode.
On the other hand the difference in CB20 was big. CB20 with 59W PPT was noticably quieter. In the 32W and 28W the fans were not spinning up at all. Kept sitting at their idle speeds.
How much the temp differences were? Not getting a noticeable result implies that the fans don't get as much air through the bottom panel as they need.
@@Juha.Latva-Somppi I said "I couldn't find a significant difference in CPU and GPU temperatures". That means the temp difference was less than 1°C. The difference could increase over time though, when the table underneath heats up and heat up the incoming air, too.
Do you reckon it would possible to add an eGPU via M.2 slot? Thanks for the detailed vid! subbed.
Maybe. The M.2 slot for the PCIe 3.0 SSD is directly connected to the CPU. I wouldn't really bother with *doing* it, but I'd sure like to see someone else try. We could even lend out a GPU if it's in the EU. // Tom
Erstmal, vielen Dank für diesen ausführlichen Review! Wie sieht es aus mit der Akkulaufzeit? Wie viele Stunden könnte man mit nur Web browsen, Code Editing, usw. erwarten?
Gerne! Ich habe die Frage so ähnlich schon dem Nutzer Shield Storm beantwortet. Dort habe ich von maximal 90 Minuten und bei leichter Last 60 Minuten geschrieben. Ich habe das alles aber nicht getestet, nur anhand der gemeldeten Leistung der Sensoren abgeschätzt. Ich war da anscheinend aber noch zu optimistisch. Die Akkulaufzeit wurde jetzt von Notebookcheck mit maximal 73 Minuten im Idle und 62 Minuten Websurfing bestimmt. Das ist wenig und ist folglich nur was für Leute mit Stromstecker in der Nähe. Die Zen 2 Desktop CPUs sind im Idle leider nicht stromsparend.
Noch ein Update dazu. Es könnte sich um Unstimmigkeiten mit dem Notebookcheck vorliegenden Notebook handeln. Sie planen einen Folgetest. Notebookcheck sagt, dass sie 60W im Idle haben, das kann ich bei meinem Apex 15 so nicht bestätigen. Bei mir sind es tatsächlich 45-48W. Könnte sein, dass die verschiedenen CPU und GPU Optionen auch unterschiedlich hohe Verbräuche haben.
Thanks for a great review! Any idea where we could find the 2060 refreshed benchmarks?
Unfortunately no. But I expect some benchmarks coming soon, f.e. on notebookcheck regarding those Refresh GPUs. I bet even inside a XMG test notebook. There are no reviews out yet because there was a NDA about the new Super and Refresh GPUs till mid April. I think this is the reason der8auer wasn't talking about the GPU inside his prototype sample at all.
hello, does you display works with nvidia drivers 531.18? I get my screen backlight off with drivers above 522.30 and newer. The screen just turns off after driver installation on apex 15, the problem is only in backlight of lcd, it even has image, what can be seen with an external light source.
Excellent review. Is the GPU soldered ? Thinking about getting one with. 2060 refresh then upgrading it to a 2070 if I ever need to. Really like that they have made this chassis tinkerer friendly.
Everything is upgradeable expect the gpu unfortunately
they soldered the GPUs onto the mainboard. Saves money, keeps pricing lower. MXM is expensive
@@TechModLab what do you mean by mxm? Is it another one of their models?
@@mudanenadaara MXM is a loosely defined standard for modular/removable GPU boards in laptops and a few small form-factor PCs. We are using it in XMG ULTRA series. ZOTAC has used it in MAGNUS EN1080K. It's not well supported by NVIDIA (and virtually not supported at all by AMD these days) and it often does not survive upgrade attempts across multiple generations due to VBIOS and heatpipe placement incompatibility. And yes, it adds a lot of R&D and validation cost while being less efficient than an integrated solution. The lack of efficiency comes from the interface of routing everything through a physical connection, but also from the limitation this puts on the general GPU board layout. // Tom
Hi! Thank you for the in-depth teardown and review!
May I ask if using this laptop purely for gaming, will there be a noticeable difference if I were to install games on the faster M2 top slot rather than on the slower M2 bottom slot?
depends on the load, but for daily tasks you will not notice it. The real benefit for gamig is minor in my opinion. But using the faster slot could help to prevent micro-stuttering.
This is boxed, wrapped and also shares the same chasis as the Acer Predator Helios. Has to be in with Acer or getting the same manufacturer to construct them. The cooling pipes look much better than the Acer Predator series. You can get an Alienware Area 51m R2 with a better screen and similar specs for about a grand cheaper than these. Great to see someone breaking in to the market, but they have a long way to go on design. Hard to charge 4k for a laptop that looks like it's from 2010.
Thank you for this very useful video. What are your thoughts after one month? And, why did you choose the 3700x instead of 3900?
New video is coming next week, will address that, too :)
In short, I still like it a lot. I spent a lot of time tuning the components, not because it is necessary, just because I like to do that. After tuning I can pretty confidently say that I was able to improve the user experience, by lowering nosie and temperature levels and keep the performance, maybe even improve it a little. Stay tuned.
I chose the 3700X because the 3900 was 150€ more expensive the day I ordered. There was a pretty noticeable price drop on the 3900 after two weeks. But still I am not unhappy. The 3700X seems to be a very good allrounder. I noticed in some reviews that the 3900 might run a little hotter and consumes more power as the 3700X in general. But I got no 3900 to verify that, impossible to clear that up. On the other hand I plan to upgrade to Ryzen 4000 'Vermeer' Desktop next year anyway, so I was looking for the best bang for the buck.
@@TechModLab thanks, very clear. After your impressions about the 3900 (despite you didn't try it) I understand that you don't recommend the 3950x at all,right? Maybe at this point if someone is looking for a very powerful laptop similar to a desktop the XMG Ultra could be is best choice ?
@@marcolondero Didn't say that. The 3950X can be a very different story. In fact the 3900 and 3900X is kind of a "failed" 3950X and is recycled. It could be - not sure just thinking here - that the "silicon quality" of the 3950X is better and it runs colder at the same performance level as the 3900, simply because it got the more efficient and more quality CCDs. In case you wonder, CCDs are the individual Chiplets AMD uses on their Ryzen CPUs. 3600 and 3700X got only one CCD, 3900 and 3950X got two. Threadripper uses up to 8 CCDs in their 3990X. Each CCD is tested after production and gets a quality label. Lower quality CCDs end up in the 3100, 3600 processors. Highest quality ones in 3950X. The other CPUs are in between. Lower quality = less max clock capable of maintaining stable, or uses more power than others for same performance.
The only thing which hold the Apex 15 back from being a "very powerful laptop similar to a desktop" as you say, is the RTX 2070. The CPU is superior to the Intel ones, not in gaming, okay, but in general they are. The Apex 15 is more lightweight and more efficient than the latest Ultra series. The new Intel 10th gen processors get hot AF. And the power supply of the Ultra series is way bulkier than the one used by the Apex 15. Hard to compare both. The Apex 15 formfactor and weight is more a real laptop than a desktop replacement as the Ultra series.
But I think the Apex 15 can easily keep up with the latest XMG Ultra 15 with a RTX 2070 inside.
@@TechModLab Everything's clear. Thank you again, sir.
Hi, thanks a lot for the review! I am usually working on higher performance (3D modelling and rendering) and mostly there are also other people in the office. Do you think the noise levels of Apex 15 are okay, or could they be disturbing for those not wearing headphones? Is it much different from other laptops? Thx!
Depends on the load of course. Hard to say to be honest. The notebook itself can get very loud if you max out the components, f.e. CAD production render (more than 51dbA possibly, i got no noise meter). But you can work in lower power modes in office (using ControlCenter or the Hotkey Fn+3). As shown in the video there are some clear sweet spots between performance and power dissipation (16:30).
XMG stated in the german forum that they are planning to lower the entertainment mode's PPT down to 45W. I expect very good results around 4000 CB20 points from that mode while maintaining the notebook quiet. Let's wait and see how it works out. :)
It gets REALLY loud if needed to cool itself use carbonaut/cryonaut on cpu and GPU as that keeps my temps low, my laptop idles in the mid to upper 40s with a pretty much silent fan curve until both cpu/gpu get in the 70s
@@sbeezynukka thanks for sharing. so you're saying it's pretty quiet up till 70 Celcius? Would you be able to measure noise levels (e.g. using a noise meter app?) And let us know? Particularly keen to know how "loud" the quiet mode is and what performance you get at that level.
Nice review. Couple of tips: Try locking your cpu at a fixed speed at lowest voltage possible. From testing with a desktop 3600 I found that the following is 100% stable: 3.6GHz@0.97V, 3.8@1.04, 4.0@1.13 which made idletemps much lower and more stable, powerconsumption dropped about 20% at 4GHz, 30% at 3.8GHz and 40% at 3.6GHz. As for ramtuning I bet you can run your ram at 3200cl18, try lowering trfc to 280-330 if it's B-die, that will improve gamingperformance a lot :)
Sounds interesting! You mean fix all cores to the same frequency in manual mode? There is a chance that it will result in less performance than just lowering the PPT. There was an interesting investigation made by pcGamesHardware, a german website. Do a quick websearch for "Ryzen-3000-Package-Power-Tracking-statt-Undervolting-1295054", you should see their website as first result.
Yes, I think I could run my RAM at 3200cl18, but we are limited to the default 1.2V SO-DIMM voltage at the moment. There is no DRAM voltage setting in BIOS yet. I maxed out the Corsair Vengeance 2666cl18 (18-19-19-19-39-58) at 3000MHz (1500MHz mclk), but it went unstable in Prime and memtest after 1-2 minutes. It was fine for gaming and making this video though.
I got my hands on some faster 2x16GB G.Skill Ripjaws 3200cl18 rated for (18-18-18-18-43-61) at 1.2V using XMP, which is great, much better than the Ballistix 3200cl16 der8auer used, they are rated for 1.35V. Unfortunately my G.Skill 3200cl18 are not stable above 3000MHz. 3000MHzcl18 (18-18-18-18-39-58) is running like a charm. They are based on Hynix C-dies, so I am stuck on some quite slow Trfc sitting at 560. 525 was unstable unfortunately.
I reached the point of maxing out the 1.2V DRAM voltage. Let's see what is possible with a tuning BIOS and more DRAM voltage.
And I will keep an eye on the upcoming Samsung 3200cl19 1.2V SO-DIMM. They could provide better timings, or worse. Relax and wait.
Cheers
@@TechModLab Yes. You will lose a bit of single core performance, but in general perf will be similar since all core run close to the same and thermals/noise is much better :) One guy I talked to ran his 3950X@0.95V which in his case improved performance and lowered thermals :) For the 3700X I would test and find a frequency/voltage at which you are satisfied with perf and nouse/thermals :) Also undervolting the GPU with afterburner and frequency-curve you can improve thermals and noise quite a bit at same performance.
@@TechModLab I have tested a lot of PPT-limit vs fixed core vs undervolt. Undervolt is barely better than stock, not worth ir, PPT is very good if you like high single score, fixed freq/voltage is good if you can sacrifice a bit ig single core for better multicore. Thermals/consumption is best at fixed freq/voltage at 3.8GHz or lower, PPT is very good around 50-70W, below 50W the frequency drpps hard. The entertainmentmode is very good in that respect, your graphs show that too :) I tested 3.8GHz@1.04 vs 65W PPT in Cinebench15: 65W PPT: 193S/1480M 54C 67W avg 3750MHz avg. 3.8@1.04V: 183S/1510M 60W avg, 51C. On my setup single score drop by 5%, multicore increase by 2%, consumption drops by 11% and 3C tempdrop. Depending on your scenario, PPT or fixed freq is best, stock or UV is not worth it.
Update. G.Skill F43200C18S-16GRS are stable at 3200Cl18 now. Used the following timings: Tcas 18, Trp 19, Trrc 18 Twrc 18, Trc 61, Tras 43, tcas write 18, trfc 600, trfc2 446, trfc4 274, tfaw 36, trrds 6, trrdl 8, twr 24, twtrs 4, twtrl 12, tcke 7, trtp 10, trdrdsc 1 trdrdscl 5, trdrdsd 5, trcpage 0, trdrddd 5, twrwrsc 1, twrwrscl 5, twrwrsd 7, twrwrdd 7, trdwr 9, twrrd 3
CmdRate 1T, GearDownEn Enabled, procODT 60
RttNom off, Rttwr RZQ/3, Rttpark RZQ/1
CAD 24,20,20,24
Voltage 1.2V
Adding it to video description, too.
@@TechModLab nice :) Any performancegains?
Nice vid. Question: why did you choose the 2666 Mhz DDR4 RAM instead of the Crucial 3200 Mhz? Does it really improve performance having a lower CAS latency?
Because I didn't like any of the available options when I ordered. So I chose the cheapest. I planned to order my own faster 32GB kit some days later. I didn't like them because the max available CAS latency was 22 for the 3200MHz one (this 22 is the CAS number for example. Just one of the important ones. There are many.)
Yes, those timings are important and can have a great effect on speed, because it determines how fast the CPU can save and aquire data from DRAM. It is not easy to give a rule of thumb here, because there are more effects to take into account than just raw memory timings. But let's say 3200Cl22 is half as fast as 3200Cl11 (Cl = CAS latency), if all the other timings would be halfed, too. So "real" 3200Cl18 memory could be up to 23% faster than 3200Cl22. That does NOT mean that your cinebench scores or in-game FPS will be 23% higher! It strongly depends on how much data is transfered between CPU and DRAM. The more data needs to be transferred the greater the impact. Cinebench for example got a quite low impact.
The other factor which is really beneficial for speed is to use an higher DRAM frequency, because of the infinity fabric clock. It is the clock that is responsible for the communication between the IO die and cpu core chiplets inside the Ryzen 3000 processor. The higher the DRAM frequency the faster the CPU's connection to "the outside world".
To make things short, in SOTTR high settings you can expect 1-4% less performance when using 3200CL22 instead of 3200cl18, maybe more on minimum FPS which are a big deal for possible micro stuttering. NOT TESTED though! Maybe I will make a followup video about that.
That answer was long... Hope that was not to much information. :)
You can also do a quick search on youtube about "ryzen 3000 ram speed". I can recommend GamersNexus and linustechtips' videos about that topic.
Cheers
@@TechModLab I'm looking for some faster ram on some website for this laptop too but all I got are 3200MHz with cl20 or cl22 , may I know what ram you are going to use?
I did quite a bit of research about this myself and found this really extensive polish benchmark that shows how much more important frequency is compared to latency:
www.purepc.pl/pamieci_ram/test_pamieci_do_procesorow_amd_ryzen_taktowania_vs_opoznienia?page=0,16
@@WenYSiow I am using G.Skill SO-DIMMs now, 2x16GB ones (G.Skill F4-3200C18S-16GRS). Unfortunately they seem to be out of stock at the moment. But there are still 8GB sticks available, if you think 16GB total is enough. There are also 3200CL16 8GB sticks. But my 16GB sticks originally running advertised with 3200CL18 are capable of running CL16, too. There is good overclocking headroom if you get lucky and get some well binned memory dies.
@@TechModLab Thx !
Thanks for the video, it was very interesting. I've just received my Apex 15 and I must say that I'm loving this laptop. However this little feller is making a huge noise out of its fans. It stays on 60-65 degrees (CPU) on idle and then it reaches 85 - 90 degrees while I start playing some games (performance mode) . Is it normal? Any suggestion on what shall I do? When I pick the quiet or balance mode in the configuration tools, the fans actually start spinning a bit lower for the sake of my ears, but the temperatures are still around 85-90 degress while playing some games.
Nothing to say about the performances, on the other hand: it's beast
Hi! Sorry for the late response, UA-cam does not show up all comments in the comment overview.
Well you can undervolt the CPU a little, you can lower the temperatures by quite a lot ding so. Unfortunately the fan noise is not really affected by doing this, because the GPU fan is the noisy part. Unless you can not limit your GPU load it will be noisy. Sorry. A good tip is to limit your FPS to a resonable value, like 60FPS, if you're willing to do so. I can understand peeps want 144FPS to enjoy 144Hz monitor.
Check out my new video. I explain CPU overclocking/undervolting there in more detail
Thx for this review, which is better 3700x with 2070 or 3900x with 2060? Is the. 2070 much more better than 2060?
depends on your use case. For gaming the 3700X and RTX 2070 is better. For content creators, video editing, rendering more CPU cores are important. But I would always recommend to take the biggest GPU possible, cause it is not upgradeable. The CPU is though
Thanks for the video, now I can compare 4800h vs 3700x in laptop....
New video is coming next week! Preparing some content concerning new BIOS' powermodes, RAM tuning (2400MHz to 3400MHz), GPU and CPU OC with undervolting, fan noise and temperatures. Also I want to address some commonly asked questions and missinformation sourrounding the Apex 15 and Clevo NH57AF1.
Really looking forward for that. I have the same laptop and it's a toaster. While in power saving mode everything is fine, but once I switch to Performance, I get Ryzen Master temperatures around 85C on Forza Horizon 4 (Entertainment mode is even worse).
So you think the 3900x is worthed?
@@stefanthis1258 Do you have the same CPU, the Ryzen 3700X? Do you have an original XMG, or a different vendor? Which BIOS version?
As a little check what could be wrong you can install HWInfo, run in Sensor-only mode, rescale the window till you see the 4th column "average" value. Now start your game and reset the values by clicking on the clock symbol down below.
Now play at least for a couple of minutes and report the average on the following values: CPU (Tctl/Tdie), CPU Die (average), CPU Package Power, GPU Temperature, GPU Power.
Maybe we can find a clue together
@@DavideStiff Hard to say, I did not get my hands on. Performance wise yes, but I am not sure about thermals. As far as I read it could be that the 3900 runs a little hotter than the 3700X. Depends on what you do with the Apex 15. Rendering a lot? 3900 or 3950X! For gaming buy the Apex 15 together with a 3600 or 3700X.
Sorry for the delay. Video will come in 1 or 2 days. Too many data and too many tests. Cutting down the content into two or three videos. Good thing is I already got most o f the data. Big respect to all those tech channels which test components all day long, it is super time consuming.
Relax and wait guys, it will be done. ;)
Bought it finally after having a rather flaky experience with a Core 15 TongFang barebone (not from XMG but the same barebone)...
Now I see how bad the battery life really is...
I wish there are denser battery (same package, voltage, but with larger capacity) made for this chassis!
Physically not possible. Maybe +30% at best with some high quality Li-ion cells, but there is no better battery technology available yet. The main problem isn*'t the battery capacity to be honest, it's the base power consumption. Desktop components are not made for power saving tasks.
@@TechModLab Yeah I see now. After looking around at one of those 160wH massive AC output camping batteries, even that will probably only last 3-4h with the laptop sipping 50W doing light tasks. And the price on those huge battery units also climb up real quick as you go up in capacity (plus the weight and the size gets impractical too).
So this config vs. the 16 inch Macbook Pro i9 8 core 2.3GHz and Radeon 5500M 8GB. Could probably beat the MBP by at least 20% I reckon? Also would you advice using it in clamshell mode to power up a desk setup with dual mons for video editing? Im thinking of upgrading from an old desktop rig that runs i6700 and GTX970.
As far as I can tell the video editing experience is flawless. I got a NVMe though, guess SATA would be different. Got a 4770K for video editng before myself. The Apex 15 is way faster. Playing back timeline multi-layered effects on 1080p 60fps video are no problem.
The Apex 15 beats the MBP with i9 by about 33% in CPU heavy stuff like Rendering, the RTX 2070 GPU is roughly 120% faster than the Radeon 5500M Pro.
Clamshell means lid closed I guess? Lid closed with two external monitors works fine. You can even use up to three external monitors using a Type-C to DisplayPort adaptor (HDMI, DP mini, USB-C /w DP). And the internal monitor as a fourth one. ;)
The bang for the buck is definitely better when you choose the Apex 15. Apple is a brand that always charges premium prices for their name. The components of the MacBooks are pretty average. The only question you have to ask yourself: do you want a higher quality case and a quieter laptop, then choose the MacBook. If you want desktop performance and an overall faster experience, choose the Apex 15.
Hope I could help you!
3DAndStuff Very nice! Thank you for the thorough reply! :)
I just bought this from Origin PC in America with a 3900x, how are you liking it so far?
Love it! After some tweaks even more. New video is coming to address that. Stay tuned.
Did you receive your unit already? I wonder how the 3900 performes in idle, light load scenarios and games compared to the 3700X.
@@TechModLab Hi thanks for your reply, I think I'll be getting mine next week. I look forward to your updated vid
is the RTX 2060 refresh one worth it?
From where did you get those two screws for the battery? Were they in the same little plastic packaging in which the battery is too? Or were they already screwed in the two holes in the battery?
they were screwed into the battery's screwholes in the notebook already, battery was extra.
@@TechModLab Okay thank you
iirc i had to take mine out the package but I have the barebones version.
@@TechModLabHave you bought it officially or did you get a demonstration version of the notebook?
@@tomik499 i am a normal customer. :) Bought mine at day 1
This one or an Omen?
@3dandstuff hey have you gotten the CC 3 2.24 update yet?
Yes, but I don't know what has changed. Anything like before.
@@TechModLab Apparantly, the change is mostly in the Active-state power management (ASPM) but the ODM says they are still working on further fine-tuning. // Tom
hello, would using a ryzen 5 solve a lot of the overheating issues?
Can't tell yet. Receiving more CPUs to test in the following days. Stay tuned, there will be another video about that.
But I am a bit confused about your statement, there are no overheating issues? What do you mean?
Kind regards :)
What are the clock speeds of the RTX 2060 refresh for the Apex? It should be the same right? Because both have 115W power limit
I don't know. But different chips, different behavior, even if the clock speeds are the same.
For example, a 10 year old processor sitting at 4GHz does NOT perform the same as a modern 4GHz processor with the same core count. The IPC are different. 2060's IPC are lower than the 2070 one
@@TechModLab What do you think how much performance difference is between the tow?
@@tomik499 Notebookcheck wrote an article back in april about that topic. They had not a lot of benchmarks till then, but they talk about an overall improvement of 7% for both, the 2060 refresh and 2070 refresh compared to the older regular ones. The RTX 2070 super is another 6-8% ahead of the regular 2070 refresh.
I was looking to buy a Sager version of this with the Ryzen 5 3600, is the noise is as bad as Notebookcheck made it out to be? I understand it won't be ultra book silent but because I use this in a living room, I don't want a jet engine either. How is the fan noise in entertainment mode with some undervolting during games?
It's not as bad as Notebookcheck claims it is. They used the extreme mode for "The Witcher" to kind of proof a point. Please check out our official reply in the comment section under the article. // Tom
@@xmg_gg Wow a reply from the very XMG! In any case, their 62 dB(A) is extreme, would the entertainment value be in the range of 50 dB(A) instead, or perhaps lower on more conservative ratings? The review cited did not go into much detail at all about it, as well as the video posted by der8auer.
@@mohammadashour692 Depends. If you want to have maximum performance close to a desktop, you have to take the fan noise as is. But if you want a quieter gaming experience you could switch to a lower powermode. Depending on the game you loose 0% to 5% of FPS average in entertainment mode, but the system will stay quieter.
Maybe I will make a followup video if a newer BIOS version is out.
@@TechModLab I am far from looking for a desktop class performance, I like the silent mode as you have shown but it seems a bit too close to the eco mode, I liked the suggestion to move it forward a bit, I have read XMG I believe stating that they might look into reducing the "Entertainment" profile down to 45 watts, I could be mistaken. I was looking to keep it into Entertainment and perhaps lower the clocks a little/voltage to keep it cool. I am fine even with a 10% decrease if it means it will run cooler and more quietly, that's the beauty (and curse) of having such a powerful chip in a laptop form. I have ordered my own now, but with the R5 3600, with 2933mhz ram, will see how it goes!
@@mohammadashour692 keep in mind that every notebook brand like XMG got their own fan-temperature profiles for the very same notebook. Your Sager could perform very different in concern of performance, noise and temperature.
is it worth to buy it with 2070 rtx or rather w8 for 3070 :?
XMG wrote in their public discord that review units of the Apex15's successor arrived yesterday. So I would wait a couple more weeks until the successor will be available. It comes with 3060 and 3070 options and Zen3 on a B550 based mainboard.
OR if you don't need all the GPU power you can wait for discounts on the older RTX20 model :)
"these modules are coming with 2400MHz (1200Mclk) SPD speed" - man, I really hate that manufacturers don't list this anywhere!
Does not matter. You can easily overclock them to their XMP profile printed on the modules via "Ryzen Master" (it get's applied in BIOS, no autostart in Windows needed). But yea, manufacturer's specs can be quite confusing about that.
@@TechModLab Well, it doesn't matter to you. And me, now that I've read about it here.
But a) in their FAQ, XMG says XMP just doesn't work. Nothing about being able to apply those profiles via Ryzen Master. And so unless people already have experience with that, they won't know...
And b) it's a hassle to figure out and enter all of those specifications you list in the description. Now, maybe it's not important to set all of those?
Anyway, thanks for pointing this out and I hope I'll be able to made it work on my future Apex 15 as well.
@@notthere83 you got a good point there. Those numbers can be confusing. The "unoptimized" first timings I posted in the video description should work on every G.Skill Ripjaws 3200 kit by copy and paste the values. The second timings may not, they are optimized to the limit of my modules.
@@TechModLab I'll probably just use the DRAM Calculator: www.techspot.com/downloads/7164-ryzen-dram-calculator.html
Worked well for techspot: www.techspot.com/review/1891-ryzen-memory-performance-scaling/
@@notthere83 True, used that for my tuning attempts too. But it's not super straight forward, as the DRAM calculator expects you to set 1.35V RAM voltage. But we only got 1.2V. works to some extend well anyway.
can gpu be fed more wattage for some performance boost
No. The GPU's BIOS is locked
i bought one 1 month ago but has not arrived yet, nor i have a date for it to be delivered.
I guess Covid-19 to blame here. Ask the support. They mentioned something about delivery shortages for non-german keyboards.
@@TechModLab thanks for the information
@@aormonde Your order hinged on an unexpected shortage in non-German keyboards which has been resolved this week. Your order should go into production early next week. Sorry about the delay! // Tom
@@xmg_gg Thank you for your sincerity.
what is the result of gaming performance in ultra high setting for SOTTR?
SOTTR Benchmark gives me the following numbers...
96 FPS (Preset "Highest", 1920x1080, Antialiasing TAA).
Min - Max - Average - 95%tile
100 - 173 - 130 - 101 (CPU Game)
106 - 252 - 152 - 113 (CPU Render)
73 - 159 - 98 - 82 (GPU)
There is no ultra high setting, "highest" only
@@TechModLab oh, sounds great. how is the result of other games? like assassin's creed odyssey and the witcher 3
@@cityhunterpro I don't own Assassin's Creed Odyssey and can't run the benchmark. The witcher 3 got no official benchmark. The only possibility to PROPERLY compare my witcher 3 FPS with other reviews is when I use EXACTLY the same scene/gameplay they used. I can't, because I don't know which part of the game they tested exactly.
still happy with XMG P507 .... its like 20 percent slower than this machine :P
The 3700x is in eco mode lile the 3950x?
Nope. The 3700X is running in the same power configuration as in desktop systems. 65 watt
Got a chance to compare different CPUs inside the XMG Apex 15.
Which benchmarks you would love to see?
Performance based on some synthetic tests and real life applications, thermals, fan noise, is there a bottleneck between GPU and CPU?. Thaks for your effort in creating a fantastic series of reviews. Waiting for CPU comparison before order one 🙂. Btw is there a BIOS options to reduce gpu performance in order to increase battery life?
@@panzerdascal1 Thanks for the input :) Any wishes on real life applications? Youncan be sire about syntethic benchs, thermals and noise.
Bottleneck is hard to test tbh. Strongly dependend on application.
@@panzerdascal1 Thanks for your kind words! :)
And no, there is no option to decrease or limit GPU in BIOS. I would love to see a 80W or 90W switch to be honest. Would be awesome for quiet gaming sessions.
@@TechModLab too bad, would be nice to have such an option, maybe Tom from XMG can push this feature to ODM. Regarding real life application, I would be interested in of the Adobe editing apps like Photoshop, Illustrator, Lightroom. Adobe Premier for video editing or any other for this purpose. Some applications for 3d modeling like maya, 3ds Max or Blender. Also would be nice to see some code compilation benchmarks, like GCC for C/C++, or java a.s.o. Thanks again and I am looking forward to see your results.
@@panzerdascal1 added Puget Benchmark for Adobe Premiere to my tests. But no other new ones. Testing takes some days already, don't want to extend it too much ':-D
So the test will contain: Cinebench R20, Puget Benchmark for Adobe Premiere, SOTTR, comparison of idle vs. videostreaming situations with noise measurements, PPT and temperatures.
Processors tested are: 3600, 3700x, 3800x, 3900, 3900x, 3950x.
As far as I can tell the 2 CCD processor (3900, 3900x, 3950x) use substantially more power all the time. We are talking about more than 10W more compared to 1 CCD processors (3600, 3700x, 3800x). May not sound much but it's audible in louder fan noise and higher temperatures.
All in all I would say: stick with 1 CCD processors if you don't need to run production renderers/editors and need all those cores. Makes the Apex 15 much more pleasant. Anything above 3700x/3800x makes no sense for gamers inside this notebook barebone at all. You just lower your enjoyment.
Schenker website Sep 29 2020 has NO info on amd laptops like this seems only bestware integrator is making the XMG modsls ?>???? Confusing as hell.
Also will they have a US power 60hz version and can their boards accept a newer rtx 3070 card in months ahead?? I want the new card in my ;aptop no rtx 2070-80 ser
bestware is their official centralized shop for all brands: Schenker, XMG and Tuxedo.
There is an US AC adaptor plug available.
The Apex 15 got no upgradeable GPU in MXM format. It's soldered. If the rtx 30 series will come to notebooks they will introduce a new model together with native Zen 3 support.
Can I add hdd?
if it fits into the sata slot, yes
@@TechModLab let's say I bought the 3600cpu and I wanted to upgrade to 3950x later on, should I change my power brick as well or can use back the original brick? Thanks for answering mate 👍
@@arnokaizer2816 you can use the original one. It's suitable for all configurations
Whery good
With your camera guidance and cutting I get sick.
Guess motion sickness is a thing :) Sorry. Tty to improve it ;)
@@TechModLab ZeosPandera has the same style. I have no problem with the head mounted view. I think the problem is the to fast cuts in between.
New XMG Apex 15 videos are uploaded!
1) CPU Overclocking & Undervolting - reduce temperatures and increase performance
ua-cam.com/video/FtLzR-i8XmA/v-deo.html
I address CPU OverClocking/Undervolting and try to improve the temperatures. Some of you observed pretty high ones.
2) Memory Overclocking & Tuning - watch how to achieve 3600MHz memory clockspeed & tight memory timings
ua-cam.com/video/qCkNcZ6WEQQ/v-deo.html # part 1: results & theory
ua-cam.com/video/RWVm1_JCTEo/v-deo.html # part 2: memory overclocking guide
Upcoming Apex 15 stuff: GPU tuning, FAQ, temperature, noise, CPU showdown (3600 up to 3950X inside the XMG Apex 15)
Cheers
bitte trainere doch dein „th“. Es klingt wirklich nicht gut im englischen!
haha, ok werde ich machen. Soll ich für das "th" statt "d" lieber "s" sagen? ;)