Sempron 2800+ versus the mighty Pentium 4!
Вставка
- Опубліковано 4 вер 2024
- ⌛ I’m back to 2004, and for absolutely NO reasons under the Sun, I’ve decided to test a Pentium 4 versus an AMD Sempron, what can possibly go wrong?
🌓 Join the UA-cam channel for extra content! / @midnightgeek99
🌔 ...or join Patreon! / midnightgeek
▶ / discord
✅ Hardware used
→ Pentium 4 2.8 Northwood & Asus P4V800-X
→ Sempron 2800+ & Albatron KX600S
→ GeForce 7600 GS
→ Radeon 9250 & Radeon 9600 XT
#retro #computer
Correction: The displayed results are correct for SuperPI, the audio is wrong...don't listen to that geek, usually he doesn't know what he's talking about!
This is a nostalgic CPU. I remember that Intel's Pentium 4 had good performance at the time, so Willamette, Northwood, and Prescott were released in the Pentium 4 series. And Zalman is a company in my country, and at the time they were famous for making really good coolers.
Bro, I enjoyed watching the video today too. thank you.
(This comment was written through a translator, so the meaning may not be accurate.)
You forgot the Cedar Mill core which was a 65nm P4. Greetings to Korea, I like your country :)
Especially Northwood was decent. The tasks where NetBurst shined were really fast, and Northwood had the high clocks of Prescott, but the higher IPC.
@@HappyBeezerStudios I agree.
@@RuruFIN thx bro.
Wow, I did not know that Zalman is from South Korea, amazing!
Zalaman are one of my favorites, I have some epic coolers from them, and also a nice case.
HT was the exact reason I switched from AMD to Intel back in the day. Because at work they had just replaced all the test machines with HT P4s, so I knew first hand what difference those made for the OS response time when one thread fully loaded a logical CPU. And man these HT P4's triggered hundreds of crazy multithreading bugs in our software. So I owe a part of my learning experience to P4 Hyperthreading.
I would've loved to see those bugs generated by the brand new HT technology. But yeah, HT was quite nice...case in point, it's still alive and kicking 22 years later.
@@MidnightGeek99 VMs weren't a thing yet and real dual CPU SMP systems were expensive. Our test only had a handful of them. But with HT, suddenly hundreds of test machines had a good chance to trigger those bugs. And they did.
I was still using a P4 3.0 Ghz Northwood with HT up until 2015 as my main PC, and I can assure you that HT was 100% useless, it did not improve performance or responsiveness at all, and in many games (even multithreaded ones) it hurt performance a little and always caused idle and load temps to jump to about 8-10º higher. Best case scenario was 7zip and Cinebench, with HT enabled it got like 3% to 5% extra performance but all other tests showed reduced or equal performance. HT did improve in the Prescott core, but that bad experience is the reason even to this day I avoid HT enabled CPU's and would rather save 30-50$ to get a 6C/6T CPU rather than a 6C/12T
I love the zip tie to the socket 478 heatsink retention bracket. Using a classic cooler there. It's after my own heart... ties and all. +1
The cooler is amazing, and yeah...I had to improvise :|
Excellent video my friend! I had almost the same exact setup back in the day. I originally had P4 2.0 Ghz and then upgraded to the Northwood 2.8 Ghz later and got a XFX 7600 GT XXX edition back in 2006. Far Cry maxed out with HDR and the latest version of Pixel shader with the latest Far Cry patch I believe it was 1.33 was the first game I tried when getting the 7600 GT. Far Cry looked stunning with HDR at 1024 x 768 which I believe the resolution I was playing at. I always thought the Athlon XP 2800+ was the better performing CPU. Good to know I got my moneys worth with the better performing CPU. So I didn’t exactly get ripped off. Didn’t know XP 2800+ didn’t support HT. Brought back memories. Great video!
The 7600 GT was an amazing card, and yes, 2.8 Northwood was quite good.
Great comparison, i like your videos.
Thank you :)
The athlon xp would have been a better pick, a 2800+ costs like $15 at most and has 512kb L2. The athlon xp was actually very competitive with pentium 4 specially oced, which was what amd recommended and marketed.
Exactly my point, Sempron was almost useless in 2004, you could have a Barton for a small price difference :)
50% load doesn't mean it can do extra stuff. It actually doesn't mean anything without knowing what exactly the thread is doing. If one thread is able to utilize all the execution units in the CPU, there is no sense in trying to load the second one, because you're already using the CPU at its max. Threads will just share execution units, slowing each other. Yes, you can get a better latency, but the throughput will stay the same or even worse. This is the reason Intel is not using HT in E-cores right now: they are simply fully utilized by one thread almost all the time. So, whether or not you could actually benefit from the HT depended on what were you doing.
This topic deserves a separate video, with proper testing :)
@@MidnightGeek99 I'm more from engineering and dev world, so the benchmarks at my job are different, but in this case the easiest way to check whether the CPU benefits from the HT is to disable it and see what happens. Still, you cannot tell how much this workload satisfies the core of the non-HT CPU and if it could benefit from the HT. Even if the OS tells it is 100% utilized, it doesn't mean it cannot theoretically do more if the workload is better optimized for this exact CPU. Different CPUs are differently designed, so as the workloads, so usually...it depends :)
As owner of a Pentium 4 HT 631 and a Celeron 440 I can absolutely confirm the results.
Your Sempron has 68% of the clock of your P4. My Celeron has 66% the clock of my P4. Prescott is slower per clock than Northwood.
The Pentium 4 can levarage SSE2, while the Sempron has to do with MMX, SSE and 3DNow!, so float heavy tasks that make use of SSE2 instructions can run much faster. Quake 3 is a good example.
Stuff that keeps the pipeline filled, like video rendering, also did well on it.
But looking at price, the P4 is maybe 10-20% faster for a much higher price. Putting an Athlon XP 2800+ against it will bridge that gap easily and would probably be a better comparison. The 3000+ is where they fell off and the 3200+ can't compete with a 3.2 GHz P4, but at that point we get the Athlon 64
For the $350 battle it would be interesting to see Intel chipsets on AMD and nvidia chipsets on AMD and how much of a different that makes. And maybe an Athlon XP as well.
Yes, Pentium 4 was more expensive, that why many of us had AMD back then. I did an Athlon XP vs P4 comparison in the past, with nforce2 and intel p865 chipsets, but the video is old and not that good...I need to make another video soon.
Die length always gives Thorton/Barton away.
Windows XP feels more responsive on my XP 3000+ than it does on a Pentium 4 HT 3.2GHz (although neither come close to dual PIII 1000).
However due to the laptop's board the P4 only has single channel DDR333 (actually have benchmarks in a spreadsheet of the performance measured by Aida64) and an older 80GB 2.5" IDE HDD instead of DDR400+250GB 2.5" SATAII HDD.
Only for the trained eye :)
Pentium 4 really needs that dual channel bandwitdh...maybe I should do some tests, single channel vs dual.
@@MidnightGeek99 It would be good to see if older/lower end single channel boards are worth considering.
K7 Semprons are cool as they're not cut-down cores (those Thortons which is are Barton 256k), only slower-clocked versions of their Athlon XP big brothers.
Yes indeed!
Upgrading from a K6-2 to a Palermo Sempron in 2012 (yes, really) was a night and day difference. The supposedly "equivalent" Celerons at mum's work ran way worse.
I bet...wow, 2012, now that's a hefty upgrade, fiat to ferrari.
Not bad, not bad. I never had a Sempron i had a Duron 1.2GHz with Riva Vanta 16Mb. This was the best gaming pc for me, the BEST. I even played Half Life 2 on it. In my green times such low end pc was better than my late multi core machine with 1Gb videocard. By the way river is beautiful
Half-Life 2 was amazing, it allowed us to use lower directx versions.
Yes, I grew up near this river, it was amazing.
Hey I would say that you forgot to compare the 400 FSB ones and 533 FSB ones, the 800
FSB Northwood is Socket 754 territory I would say.
Yea I was thinking the same. A 400fsb P4 would have been the better choice here for the test.
I didn't forgot, I just did not test them :))
In the future I will test more CPUs, and maybe add the results together.
The Barton Semprons are a bit better, lower wattage than the XP Bartons and same L2 cache. Those would be the Sempron 3000+ and 3300+.
Yes, XP Bartons are the bottom of the barrel, celeron-like level of performance?
0:39 If it is 2004, then why not go with Sempron 3100+ s754 and NForce 3 250 board? It still should be cheaper than P4 combo at that time and Q3 really likes s754/939. ;)
I needed a random premise for the video, in order to make the video :))
I have a s754 Sempron, I'm going to give it a spin in the future, don't worry.
Oh.... Battle of the Titans
Indeed :))
Unde ai găsit procesoarele și plăcile? Sunt extrem de greu de găsit, într-o condiție cât de cât ok.
OLX :)
Socket A Sempron were truly amazing price and performance was on point, Socket 939 ones not so much.
Great video, and very comprehensive testing
10x! Sempron was ok, its main issue was that Athlon XPs were also cheap :)
@@MidnightGeek99
Yeah but Duron were even cheaper.
I had Duron 1.6 GHZ on crappy via chipset motherboard and upgrading from P-II 333mhz, it was amazing, playing Quake 3 i even try out the FarCry but that didn't go as well on GF 4 MX.
Latter i upgraded to Athlon XP 3200+ barton core and that was rock solid machine on nvidia chipset motherboard.
A7N8X-Deluxe, best motherboard for socket A
Amd Sempron sau Duron erau echivalentul la Intel Celeron. In acea perioada am mers pe AMD Athlon SH pt. ca era mult mai ieftin si avea tehnologia 3DNow! dar aveau o mica, mare problema, daca iti aduci aminte, procesoarele AMD, fiind ieftine, nu aveau capacul de deasupra nucleului precum cele de la Intel si majoritatea celor SH erau ciobite din cauza mainpularii cooler-ului. Baga si un review cu proceasoare AMD ciobite de care toata lumea fugea, ne uitam la ele cu lupa si nu stiu daca este mit sau realitate, cum ca cele ciobite erau terminate si dadeau erori.
Eu unul nu am avut Athloane ciobite...pot sa ciobesc unul daca vrei :))
why Sempron and not Athlon XP?
K7 Sempron is practically an Athlon XP, they're not castrated like Celerons are.
He probably just had that Sempron around.
But that means we can have a Celeron vs Athlon XP run later.
And perhaps even compare those results to the ones in this video for the ultimate 2.8 GHz overview.
I already did an Athlon XP vs P4 video before, and I wanted to see Sempron's capabilities.
@HappyBeezerStudios Neah, I was just curious. But yes, I will need to add an XP and Celeron into the mix in the future :)
Accentul ala de roman:)))
Trebuie sa iesim cumva in evidenta, nu? :D