@@xkais487 well the 5800x will probably clock a bit higher, but I don't think it has any other advantages. Depends on whether you are an overclocker or not. Those are great prices anyway though. Remember how much a 9900k was new haha? Good times either way you go
1:26 Honestly pretty impressive pulling ~180W average and 240W max and peak at 56C(68C). I think you could under-voltage the card and maybe cut down the power by 30W without loosing much performance. How fast is the 7800 XT, about the same speed as the RTX 4070? Makes me wonder how much power the RTX 4070 would use under this load.🤔 Do Cinebench R24 used optix on RTX cards?
It is a good card for sure. Speed comparison depends on workload. In blender rendering, the Nvidia GPUs are much faster. Gaming, raster work is much closer. Cinebench uses HIP for AMD and CUDA for Nvidia. I can run the 4070 to compare in cinebench. I will test all of my GPUs on it eventually. Maybe do some on Livestream
I tried google if cinebench uses HIP-RT and Optix. I did not find a good answer but it looks like it uses Optix and HIP (non-RT). But I am not 100% sure. But in that case the RTX 4070 would probably be a lot faster.
@@ContradictionDesign heh you are quick to assign more work for yourself. 😄 As far a i understand it cinebench is a benchmark tool made by Maxon, that makes Cinema 4D and I think the render engine in Cinema 4D is called Redshift, and in the Redshift 3.5.15 Release Notes They say they are using Optix 7.4 On nvidia cards. So I would expect that to be the case for the benchmark too. But I cannot find it using google or on there home page. As far as I understand it Optix uses the RT cores of the GPU, and cuda uses the cuda cores, that are simple computation cores and are the part of the GPU used for rasterisation when playing non RT games. For AMD I think HIP is the same as cuda and HIP-RT is using the RT cores in the GPU. But honestly I am not sure this is correct, but it is how I understand it. I don't know if "task manager" (windows) can show graphs for RT vs compute vs 3D, but maybe one would be able to see what part of the GPU Cinebench R24 is using. I just think it is most fair to compare RT cores vs RT cores and compute cores vs compute cores. Not that it matters if you cannot use the RT cores the fastest GPU is the fastest no matter what they call the part of the GPU that does the work. But if we are to generalize the results to other software's. (as you say, it depends on workload) But all this is out of scope for your video, so sorry for bringing it up. The video shows that the cooler on that card is grate and that is nice to know in case someone sees a good black Friday deal on that card 👍
@@Petch85 Well first off, don't ever worry about adding to the discussion here! Thanks for new ideas! When I run cinebench it says it uses CUDA on Nvidia and HIP on AMD. But as far as testing that, I don't even know if hwinfo breaks out which cores are active or not. But yeah comparing HIP-rt and Optix is the most fair if possible. That's why I like to test in Blender for RT, because it uses those cores properly. Thanks for the discussion! Love it all
I ran Cinebench GPU test on my XFX QICK RX 7800XT just a few minutes ago and on the last pass, my screen turned off and the lights on my PC went off except for the power indicator. I was tempted to force restart it but the fans were still running. So I waited and my screen and PC went back on after about 5 minutes. It did not reboot; everything was as it was before the screen turned off. By the time the screen went back on, the test was finished. But I got a really low score of just 1142! The temps were OK during the test, or at least, as HWInfo reported. Peak GPU temp during the test was ~64C. Is there something wrong with my system or GPU?
Oh no sorry to hear that. Normally screen problems like that are caused by the GPU or bad drivers/ software support. I get crashes in Blender sometimes that your issue reminded me of. It definitely doesn't sound like overheating then, since that does force the PC to shut down or reboot. I would get the newest driver and see if it happens again. Pretty weird though. Let me know if you are still having issues afterwards
I have an xfx speedster 7800 xt, tried this test, the first and second time I got less than 2000 points, the third just over 4000, and now it's reached the final square and it stopped, tried it on uningine and I'm getting around 5 fps more than a 7700 XT while it should be getting ten or 11 more. I'm quite sure the drivers are updated and there are no apps running in the background during the tests, should consider returning the gpu?
@@CanBreathe Sorry to hear that you're having trouble. Have you tried monitoring the temperature while you run a test? It reminds me of thermal throttling
@@ContradictionDesign yeah, it wasn't thermal throttling since its max temp was around 70 degrees, it has kinda fixed itself on cinebench reaching 10538, looks like it was a bug, still not reaching top perf on uningine but it's not too far off, thanks for giving reference for the stats!
@@CanBreathe Hey you're welcome. Glad to hear it's working better. There are so many factors to performance in PCs now. It is hard to diagnose issues now.
Something may be wrong with my system because I cannot even get my 7800XT to score 500. It literally scored lower than my 750Ti. My 6600XT beat it 14-fold.
Does the GPU run other apps/games as expected? Well if I were you, I would look into driver conflicts. I know that for Blender, you have to have certain drivers to run HIP-RT on AMD GPUs. So start there. But yeah that is weird behavior for sure!
@ContradictionDesign it was the driver. AMD Software: Adrenalin Edition showed that the driver was up to date, but it turned out to be 7 months out of date. It is performing properly after I manually installed the latest driver.
Was this test done on all stock settings? My temp sits around 70⁰ with hot spot peaking at 93⁰ however i undervolted a little and increased accesible power limit by 15%
@@brendan9340 yes this test was all stock settings. I normally don't overclock for rendering tests, because they can be harder to compare. But it did pretty good! Hey I am used to 105 memory hotspots so that sounds like a decent OC to me! Of course those were Nvidia cards.....
I want to take 4060 or 4060ti, I can't decide which video card will be faster in blender 4.0 and whether there is a need to overpay for 4060ti or it is better to take 4060 ProArt 8gb?
I would move up to the ti if you are going to get the 16 GB version. It is very nice to have for Blender work. But if you are settled on the 8 GB models, the 4060 ti 8 GB should be about 20% faster just in raster performance, according to techpowerup, and is likely even faster in raytraced work like rendering. So depending on the price difference, that is the performance gap you could decide with. But I don't think you would feel a lot of difference between the 4060 and 4060 ti.
Yeah I would say it is pretty normal. I never pay attention to idle temps though, so not for sure. Also, this is the smaller cooler. Do you have the same model as mine?
I have the PC red devil and mine us running around 75C just with halo infinite on high settings. Was running cooler before all the new updates. Is 75C OK to run it for couple hours ? My fans are all at normal speed. 6 fans total. Thanks
@@paulrakis Yeah it should be ok. My 3090s used to run at 65 core, but 95+ on the memory. They can handle quite high temps. Do you have a switch with eco/OC on that model?
is rx7700xt good for blender and content creation or should i go with rtx 4060ti 8gb... (i cant afford the 4070 due to the high taxation on pc hardware in my country)
Well if you need more VRAM the AMD will be better but the 4060 ti is faster for rendering with cycles. If you won't use cycles much you probably would be fine with the AMD. I haven't tested the 7700 XT though, so I don't have exact numbers for rendering yet
I do not have any of these CPUs, so it is hard for me to say. AMD R5 7000 and Intel i5 13th gen are probably very close to each other for 3D, but they are all good "low-end" CPUs to start out with. The biggest thing is that all of the modern CPUs have very high clock speeds and IPC, so most 3d workloads that rely on CPU will do well compared even to older i9 and R9 CPUs, because very few workloads are heavily-threaded. The only time you use all the threads is in simulations and rendering on CPU, but those are both fringe use cases for many people. Even simulations are not always able to use the full CPU. And even if they do, you need 64+ GB of RAM to run such large sims. So for midrange systems or systems with a great GPU and one of these midrange CPUs, the R5 and i5 series will all be very good while saving you money.
@@ContradictionDesign hhmmm so basically ur saying that any new gen CPU either its low or mid tier its already good enough for 3d. and has a great price for it too. so going with anything is actually a good deal too. fair enough. but for the AM5 socket and a nice price CPU is it actually the "better" one among the other nice-priced AM5 CPU(ryzen 5 7500f)? or is it still the same reason with the statement before?
@@njswk6843 I think very few processes use more than 8 cores regularly. But even 6 cores with high clock speed will run smooth in 3d. If you know you need higher core counts, then you know. otherwise 8 or 6 is plenty. As far as the 7500f, it looks to me that it is likely a little faster than the 5600x, due to clock speed improvements. Also, it is available for very good prices. The biggest issue is that AM5 is still buggy and has bad RAM support and slow boot times. It can be a hassle to set up an AM5 system. So be careful if you are not familiar with PC building, because you will be in the bios changing things a few times probably.
@@ContradictionDesign ive found some good youtuber here in my country already making the AM5 PC, and they do actually accept costume PC build from their viewers. so I might take that risk. there's this nice price Pc with rx 6600 and R5 7500f that they made. really infested on that one now. because ofc its already future proof for upgrade and ddr5. i can saved up for a while and change the gpu to rtx 4000s series later. thanks again for the information my guy
You would have to find clients to render scenes for. Which is difficult, because the competition is large farms that have plenty of machines available. They are expensive to use, but they can finish projects for their clients as fast as desired. So there would be huge start-up cost for you or I to build a render farm, and then it is hard to get the work. I use my farm to run my own scenes. But you could also try to join Render Network, or some of the other AI projects to try to earn money. But that is basically mining anyway. Making money from rendering is not as easy as mining, to be clear.
The hellhound is a cooling beast. Never got over 47 at 1050rpm with a hotspot of 65. Scored. 11000ish on 3 runs. The fans even turned off a few times lol
@@ContradictionDesignthought I would give an update after tuning in my overclock. 1050v 2575ram 2750core fast timings and no extra power because it only gives about 2% more performance on my card over the pure undervolt. 50c and 75 hot spot. Hot spot can get to 80 if the game is very CPU friendly since my case fans are tied to that temp. Even when I cranked it to 15% power and was pushing 2900 on the core clock it never gets over that 80c so I'm guessing the hotspot for the hellhound is the memory, some days it likes 2600 others its laughably unstable. And if I drop the memory clock back down to stock it goes back to my 65-70 hotspot depending on room temp.
@@A_Name_ Very nice. Sounds like you can really push it a bit and stay fairly cool. I am used to 105c VRAM from my 3090s haha. So 80 sounds great to me!
Pretty impressive results, with a xfx quick 7800xt with 3 fans, i got 64°c max, and 1000rpm too, so this sapphire does a great job with only 2 fans!
Yeah thanks for sharing that! This GPU is 236 watt TDP, so it probably doesn't need 3 fans. But it is crazy this 2 fan card did that well then!
@@ContradictionDesign Btw do you recommend me to get 5800x at 214€ or 5700x at 194€ there are about the same performance but the 700x uses less power
@@xkais487 well the 5800x will probably clock a bit higher, but I don't think it has any other advantages. Depends on whether you are an overclocker or not. Those are great prices anyway though. Remember how much a 9900k was new haha? Good times either way you go
In red dead redemption 2 in ultra settings , 2k resolution I have 65-66c
1:26
Honestly pretty impressive pulling ~180W average and 240W max and peak at 56C(68C).
I think you could under-voltage the card and maybe cut down the power by 30W without loosing much performance.
How fast is the 7800 XT, about the same speed as the RTX 4070? Makes me wonder how much power the RTX 4070 would use under this load.🤔
Do Cinebench R24 used optix on RTX cards?
It is a good card for sure. Speed comparison depends on workload. In blender rendering, the Nvidia GPUs are much faster. Gaming, raster work is much closer.
Cinebench uses HIP for AMD and CUDA for Nvidia. I can run the 4070 to compare in cinebench. I will test all of my GPUs on it eventually. Maybe do some on Livestream
I tried google if cinebench uses HIP-RT and Optix. I did not find a good answer but it looks like it uses Optix and HIP (non-RT). But I am not 100% sure. But in that case the RTX 4070 would probably be a lot faster.
Yeah 4070 would use CUDA. So the C24 tests might show them pretty close together. I will have to start this test series soon.
@@ContradictionDesign heh you are quick to assign more work for yourself. 😄
As far a i understand it cinebench is a benchmark tool made by Maxon, that makes Cinema 4D and I think the render engine in Cinema 4D is called Redshift, and in the Redshift 3.5.15 Release Notes They say they are using Optix 7.4 On nvidia cards. So I would expect that to be the case for the benchmark too. But I cannot find it using google or on there home page. As far as I understand it Optix uses the RT cores of the GPU, and cuda uses the cuda cores, that are simple computation cores and are the part of the GPU used for rasterisation when playing non RT games. For AMD I think HIP is the same as cuda and HIP-RT is using the RT cores in the GPU. But honestly I am not sure this is correct, but it is how I understand it.
I don't know if "task manager" (windows) can show graphs for RT vs compute vs 3D, but maybe one would be able to see what part of the GPU Cinebench R24 is using.
I just think it is most fair to compare RT cores vs RT cores and compute cores vs compute cores. Not that it matters if you cannot use the RT cores the fastest GPU is the fastest no matter what they call the part of the GPU that does the work. But if we are to generalize the results to other software's. (as you say, it depends on workload)
But all this is out of scope for your video, so sorry for bringing it up. The video shows that the cooler on that card is grate and that is nice to know in case someone sees a good black Friday deal on that card 👍
@@Petch85 Well first off, don't ever worry about adding to the discussion here! Thanks for new ideas! When I run cinebench it says it uses CUDA on Nvidia and HIP on AMD. But as far as testing that, I don't even know if hwinfo breaks out which cores are active or not. But yeah comparing HIP-rt and Optix is the most fair if possible. That's why I like to test in Blender for RT, because it uses those cores properly.
Thanks for the discussion! Love it all
Mine stsys below 70⁰celcius while gaming but but it feels hot af :( is that normal ?😊
Yeah that should be fine. You should feel my 3090s when they are on. 95c haha. It just means your cooler is working fine.
I ran Cinebench GPU test on my XFX QICK RX 7800XT just a few minutes ago and on the last pass, my screen turned off and the lights on my PC went off except for the power indicator. I was tempted to force restart it but the fans were still running. So I waited and my screen and PC went back on after about 5 minutes. It did not reboot; everything was as it was before the screen turned off.
By the time the screen went back on, the test was finished. But I got a really low score of just 1142!
The temps were OK during the test, or at least, as HWInfo reported. Peak GPU temp during the test was ~64C.
Is there something wrong with my system or GPU?
Oh no sorry to hear that. Normally screen problems like that are caused by the GPU or bad drivers/ software support. I get crashes in Blender sometimes that your issue reminded me of. It definitely doesn't sound like overheating then, since that does force the PC to shut down or reboot.
I would get the newest driver and see if it happens again. Pretty weird though. Let me know if you are still having issues afterwards
@@ContradictionDesign Thanks! I'll check if there are any driver updates.
I have an xfx speedster 7800 xt, tried this test, the first and second time I got less than 2000 points, the third just over 4000, and now it's reached the final square and it stopped, tried it on uningine and I'm getting around 5 fps more than a 7700 XT while it should be getting ten or 11 more. I'm quite sure the drivers are updated and there are no apps running in the background during the tests, should consider returning the gpu?
@@CanBreathe Sorry to hear that you're having trouble. Have you tried monitoring the temperature while you run a test? It reminds me of thermal throttling
@@ContradictionDesign yeah, it wasn't thermal throttling since its max temp was around 70 degrees, it has kinda fixed itself on cinebench reaching 10538, looks like it was a bug, still not reaching top perf on uningine but it's not too far off, thanks for giving reference for the stats!
@@CanBreathe Hey you're welcome. Glad to hear it's working better. There are so many factors to performance in PCs now. It is hard to diagnose issues now.
Something may be wrong with my system because I cannot even get my 7800XT to score 500. It literally scored lower than my 750Ti. My 6600XT beat it 14-fold.
Does the GPU run other apps/games as expected?
Well if I were you, I would look into driver conflicts. I know that for Blender, you have to have certain drivers to run HIP-RT on AMD GPUs. So start there.
But yeah that is weird behavior for sure!
@ContradictionDesign it was the driver. AMD Software: Adrenalin Edition showed that the driver was up to date, but it turned out to be 7 months out of date. It is performing properly after I manually installed the latest driver.
@@zsamps awesome! Great to hear it! Drivers are always a point of drama with GPUs haha
Was this test done on all stock settings? My temp sits around 70⁰ with hot spot peaking at 93⁰ however i undervolted a little and increased accesible power limit by 15%
@@brendan9340 yes this test was all stock settings. I normally don't overclock for rendering tests, because they can be harder to compare. But it did pretty good!
Hey I am used to 105 memory hotspots so that sounds like a decent OC to me! Of course those were Nvidia cards.....
I want to take 4060 or 4060ti, I can't decide which video card will be faster in blender 4.0 and whether there is a need to overpay for 4060ti or it is better to take 4060 ProArt 8gb?
I would move up to the ti if you are going to get the 16 GB version. It is very nice to have for Blender work. But if you are settled on the 8 GB models, the 4060 ti 8 GB should be about 20% faster just in raster performance, according to techpowerup, and is likely even faster in raytraced work like rendering. So depending on the price difference, that is the performance gap you could decide with.
But I don't think you would feel a lot of difference between the 4060 and 4060 ti.
is that normal temps for gpu junction at idle? like 54c ? mine got like 50c idle, never pass 72 on a stress test. Was wondering if its normal
Yeah I would say it is pretty normal. I never pay attention to idle temps though, so not for sure. Also, this is the smaller cooler. Do you have the same model as mine?
@@ContradictionDesign i have the powercolor hellhound 7800xt model
@@josesalvador7805 Oh nice! I bet you just have a better cooler then!
I have the PC red devil and mine us running around 75C just with halo infinite on high settings. Was running cooler before all the new updates. Is 75C OK to run it for couple hours ? My fans are all at normal speed. 6 fans total. Thanks
@@paulrakis Yeah it should be ok. My 3090s used to run at 65 core, but 95+ on the memory. They can handle quite high temps. Do you have a switch with eco/OC on that model?
is rx7700xt good for blender and content creation or should i go with rtx 4060ti 8gb... (i cant afford the 4070 due to the high taxation on pc hardware in my country)
Well if you need more VRAM the AMD will be better but the 4060 ti is faster for rendering with cycles. If you won't use cycles much you probably would be fine with the AMD. I haven't tested the 7700 XT though, so I don't have exact numbers for rendering yet
hello, came back again... now im wondering with the Ryzen 5 7500f power, is it good for 3d? better than the ryzen 5 5600x or i5 12/13 gen?
I do not have any of these CPUs, so it is hard for me to say. AMD R5 7000 and Intel i5 13th gen are probably very close to each other for 3D, but they are all good "low-end" CPUs to start out with.
The biggest thing is that all of the modern CPUs have very high clock speeds and IPC, so most 3d workloads that rely on CPU will do well compared even to older i9 and R9 CPUs, because very few workloads are heavily-threaded. The only time you use all the threads is in simulations and rendering on CPU, but those are both fringe use cases for many people. Even simulations are not always able to use the full CPU. And even if they do, you need 64+ GB of RAM to run such large sims.
So for midrange systems or systems with a great GPU and one of these midrange CPUs, the R5 and i5 series will all be very good while saving you money.
@@ContradictionDesign hhmmm so basically ur saying that any new gen CPU either its low or mid tier its already good enough for 3d. and has a great price for it too. so going with anything is actually a good deal too. fair enough.
but for the AM5 socket and a nice price CPU is it actually the "better" one among the other nice-priced AM5 CPU(ryzen 5 7500f)? or is it still the same reason with the statement before?
@@njswk6843 I think very few processes use more than 8 cores regularly. But even 6 cores with high clock speed will run smooth in 3d. If you know you need higher core counts, then you know. otherwise 8 or 6 is plenty.
As far as the 7500f, it looks to me that it is likely a little faster than the 5600x, due to clock speed improvements. Also, it is available for very good prices. The biggest issue is that AM5 is still buggy and has bad RAM support and slow boot times. It can be a hassle to set up an AM5 system. So be careful if you are not familiar with PC building, because you will be in the bios changing things a few times probably.
@@ContradictionDesign ive found some good youtuber here in my country already making the AM5 PC, and they do actually accept costume PC build from their viewers. so I might take that risk.
there's this nice price Pc with rx 6600 and R5 7500f that they made. really infested on that one now. because ofc its already future proof for upgrade and ddr5. i can saved up for a while and change the gpu to rtx 4000s series later. thanks again for the information my guy
@@njswk6843 Hey that is a great machine to start with! Good luck and you are welcome!
How can earn money from mining computer to render
You would have to find clients to render scenes for. Which is difficult, because the competition is large farms that have plenty of machines available. They are expensive to use, but they can finish projects for their clients as fast as desired. So there would be huge start-up cost for you or I to build a render farm, and then it is hard to get the work.
I use my farm to run my own scenes. But you could also try to join Render Network, or some of the other AI projects to try to earn money. But that is basically mining anyway. Making money from rendering is not as easy as mining, to be clear.
The hellhound is a cooling beast. Never got over 47 at 1050rpm with a hotspot of 65. Scored. 11000ish on 3 runs.
The fans even turned off a few times lol
Wow yeah those are very good temps.
@@ContradictionDesignthought I would give an update after tuning in my overclock.
1050v 2575ram 2750core fast timings and no extra power because it only gives about 2% more performance on my card over the pure undervolt.
50c and 75 hot spot. Hot spot can get to 80 if the game is very CPU friendly since my case fans are tied to that temp. Even when I cranked it to 15% power and was pushing 2900 on the core clock it never gets over that 80c so I'm guessing the hotspot for the hellhound is the memory, some days it likes 2600 others its laughably unstable. And if I drop the memory clock back down to stock it goes back to my 65-70 hotspot depending on room temp.
@@A_Name_ Very nice. Sounds like you can really push it a bit and stay fairly cool. I am used to 105c VRAM from my 3090s haha. So 80 sounds great to me!