Which is faster Athlon XP 2400+ or Sempron 2800+ 754?

[Sharptooth]

Originally posted by: carlosd
Which CPU will be faster at stock (no Ocing, I know that OCed the sempy will kill the Athlon XP) I cannot find any direct comparison between those CPUs?. Could you provide me some links. A semprom 2800+ socket A Vs. Semprom 2800+ S754 performance comparison would also be useful.

Even if you did overclock the socket A 2800+, it would have to be of immense disparity (MHZ and FSB) to overtake the Sempron S754 that it wouldn't be worth the effort. So in stock form for both, it's an easy answer...get the Sempron 2800+ (S754) because it'll be faster across the board. Synthetic benchmarks have some usage as a basis of comparison but that's only part of the picture. 'Real-World' apps will confirm the true nature of the beast.

--Fernando

 

[maluckey]

You don't need large OC's to catch the 754/Sempron. You need about 50-75 MHz and 3-5 MHz FSB to even it out, depending on RAM settings. Super-Pi backs this up on the scores as far as CPU, and quite a few benches on 3D mark show this.

Granted 3DMark 03 is synthetic, but to compare Sempron 754 to NF2 Barton with 6600 and 5900 cards respectively, and the scores show that the Sempron is often beaten. My own score with an overclocked 5900/NF2/Barton beats quite a few OC'ed 6600/Sempron/754 combos.


General Information
Operating System Microsoft Windows XP Microsoft Windows XP


DirectX Version 9.0c 9.0c


Mobo Manufacturer http://www.abit.com.tw/


Mobo Model NF7-S/NF7,NF7-V (nVidia-nForce2) C51MCP51


AGP Rates (Current/Available) N/A / N/A N/A / N/A


CPU AMD Athlon(tm) XP/MP/4 2418 MHz AMD Sempron(tm) 2403 MHz


FSB 209 MHz 300 MHz


Memory 1024 MB 1022 MB




Display Information
Graphics Chipset NVIDIA GeForce FX 5950 Ultra NVIDIA GeForce 6600


Driver Name NVIDIA GeForce FX 5950 Ultra NVIDIA GeForce 6600


Driver Version 7.7.7.2 8.1.9.8


Driver Status WHQL - Not FM Approved WHQL - FM Approved


Video Memory 256 MB 256 MB


Core Clock N/A 299 MHz


Memory Clock N/A 577 MHz




Sound Information
Sound Adapter Driver Name SB Audigy Audio [C000] Realtek AC97 Audio


Sound Adapter Driver Version 5.12.2.444 5.10.0.5900




Benchmark Settings
Program Version 3DMark03 Revision 6 Build 0 3DMark03 Revision 6 Build 0


Resolution 1024x768@32 bit 1024x768@32 bit


Texture Filtering Optimal Optimal


Pixel Processing / Antialiasing None None


Post processing false false


Vertex Shaders Optimal Optimal

Final scores? Guesses?

The Barton won the CPU tests 745 to 692, the Wings of Fury 200.4 to 181.7, Battle of Proxycon by 1.4 points and lost the rest by less than 1.5 FPS. The NF2/Barton/SB Audigy Gamer won all Audio tests. At 24 sounds by 4 FPS, Coincidentally, the Sempron/754/6600 won only two of the 3D, and by less than 1.5 FPS; nothing in the 3D by more than two points. That means in-game with 3D hardware sound; my obsolete rig is faster all-around

 

[dug777]

Originally posted by: maluckey

Slap a 7800 AGP card in a socket A system, and it will compete in gaming with nearly any socket 939 system, beating most (without 7800's)..

Good point!!

On the average, people with a 939 combo and a 6800 NU will lose out versus a well built Socket A and a 7800 series AGP card. PCI e is great, but the performance gap is large between vid card generations. It's unlikely that a faster mobo chipset can make that big of a difference.

clearly someone who has never gamed with a socket A recently I am CPU bound in newer games with a 6600GT, i'd hate to hobble anything higher with a socket A (mine is 'only' at 2.1Ghz admittedly )...so the bloke with a 7800 (and it would have to be a GS, being AGP and all) would be enjoying significantly lower min fps (i did a comprehensive comparison in FEAR a while back in video, comparing my skt A system to A64 rigs)...

You can just ask synth about jumping from a 2.4Ghz tbred-b to an opty with his 6800GT, the difference was night and day for him in gaming (those all important minimum fps remember ).

 

[carlosd]

Originally posted by: maluckey
You don't need large OC's to catch the 754/Sempron. You need about 50-75 MHz and 3-5 MHz FSB to even it out, depending on RAM settings. Super-Pi backs this up on the scores as far as CPU, and quite a few benches on 3D mark show this.

Granted 3DMark 03 is synthetic, but to compare Sempron 754 to NF2 Barton with 6600 and 5900 cards respectively, and the scores show that the Sempron is often beaten. My own score with an overclocked 5900/NF2/Barton beats quite a few OC'ed 6600/Sempron/754 combos.


General Information
Operating System Microsoft Windows XP Microsoft Windows XP


DirectX Version 9.0c 9.0c


Mobo Manufacturer http://www.abit.com.tw/


Mobo Model NF7-S/NF7,NF7-V (nVidia-nForce2) C51MCP51


AGP Rates (Current/Available) N/A / N/A N/A / N/A


CPU AMD Athlon(tm) XP/MP/4 2418 MHz AMD Sempron(tm) 2403 MHz


FSB 209 MHz 300 MHz


Memory 1024 MB 1022 MB




Display Information
Graphics Chipset NVIDIA GeForce FX 5950 Ultra NVIDIA GeForce 6600


Driver Name NVIDIA GeForce FX 5950 Ultra NVIDIA GeForce 6600


Driver Version 7.7.7.2 8.1.9.8


Driver Status WHQL - Not FM Approved WHQL - FM Approved


Video Memory 256 MB 256 MB


Core Clock N/A 299 MHz


Memory Clock N/A 577 MHz




Sound Information
Sound Adapter Driver Name SB Audigy Audio [C000] Realtek AC97 Audio


Sound Adapter Driver Version 5.12.2.444 5.10.0.5900




Benchmark Settings
Program Version 3DMark03 Revision 6 Build 0 3DMark03 Revision 6 Build 0


Resolution 1024x768@32 bit 1024x768@32 bit


Texture Filtering Optimal Optimal


Pixel Processing / Antialiasing None None


Post processing false false


Vertex Shaders Optimal Optimal

Final scores? Guesses?

The Barton won the CPU tests 745 to 692, the Wings of Fury 200.4 to 181.7, Battle of Proxycon by 1.4 points and lost the rest by less than 1.5 FPS. The NF2/Barton/SB Audigy Gamer won all Audio tests. At 24 sounds by 4 FPS, Coincidentally, the Sempron/754/6600 won only two of the 3D, and by less than 1.5 FPS; nothing in the 3D by more than two points. That means in-game with 3D hardware sound; my obsolete rig is faster all-around

Superpi and 3D mark are synthetic and do not reflect of the REAL performnce. Why don't you show us the difference with some real games and apps? I will show you (of course you will be also very GPU limited with DX9 games with the crappy FX5900).

In Doom 3 the semprom 2800+ (at 1.6GHz) blows away the Barton 3000+ (@ 2GHz) by 27%!!!!!!!!!. Both using the 6800GT. Given that 6600 is faster in Doom 3 than the FX5900, I can't tell how big the difference is going to be.

http://www23.tomshardware.com/cpu.html?modelx=33&model1=270&model2=266&chart=69


Let's see another game: UT2004. Here the semprom 2800+ also blows away the barton 3000+ by 27%!!!!!!!!!!! , huge!

I can't imagine what will be the difference with more demanding games like Quake 4 or battlefield 2. Here also the 6600 blows the FX5900 away.
http://www23.tomshardware.com/cpu.html?modelx=33&model1=270&model2=266&chart=71


The semprom 754 at 2.4 will blow away your computer in EVERYTHING (I doubt you could compensate a 30-40% of diference with 50-75MHz:laugh: ) except in the uselessynthetic bechmarks. See how useless are those synth benchmarks?
To catch up with 754 you will have to OC your CPU 30-40% or more, that would be like 600 or 700Mhz. I realy doubt your AXP could reach 3GHz.:laugh:

 

[ElFenix]

is the OP asking about a barton or a tbred?

 

[carlosd]

Double post

 

[carlosd]

Originally posted by: maluckey
You don't need large OC's to catch the 754/Sempron. You need about 50-75 MHz and 3-5 MHz FSB to even it out, depending on RAM settings. Super-Pi backs this up on the scores as far as CPU, and quite a few benches on 3D mark show this.

That means in-game with 3D hardware sound; my obsolete rig is faster all-around

So according to your conclusions based on an almost useless (or with limited usefulness) bechmark like 3D mark lets draw some other ridiculous conclusions:
http://www23.tomshardware.com/cpu.html?modelx=33&model1=238&model2=212&chart=60

-The P4 520 is faster in gaming that an A64 FX-57. lol!!:laugh: WTF!!!

-every processor of the 5xx and 6xx series from intel are faster in gaming than any A64 CPU. lol!!!

Those are ridiculous conclusions that go to the thrash when running real games and the A64s outperform the P4s in gaming.

Just to show how unaccurate this benchmark can be

 

[Zap]

Originally posted by: ElFenix
is the OP asking about a barton or a tbred?

The OP didn't mention whether it was a mobile or not. If desktop then a Tbred, mobile a Barton. Also, he edited his post now to say a socket A Sempron 2800+ versus a socket 754 Sempron 2800+. Both are "rated" at 2800+ by AMD. IIRC the socket A Semprons were Tbreds or Thorntons.

I'd still choose the socket 754. Even if AMD says they run close enough to deserve a similar rating, the socket 754 CPU will run cooler, be more durable with the IHS, have a larger HSF choice at this time, have the option of motherboards with PCIe and of course have the option of an AGP motherboard with the retail CPU for $79 at Fry's.

 

[maluckey]

So according to your conclusions based on an almost useless (or with limited usefulness) bechmark like 3D mark lets draw some other ridiculous conclusions:

Yup!

In 3dMark03 the CPU speed is fairly irrelevant. My best score were at 2100 Mhz. RAM speed and timings, HTT and Sheer GPU power rule that bench.

I was trying to demonstrate that for 3D apps, that you don't need the latest mobo/CPU combo, just the fastest video card around. Now you can get the nVidia 7800 series for NF2 I think I'll stay awhile with my favorite board.

carlosd,

You also pointed to the THG sites info on UT2004. It shows 1.6 FPS in a CPU dependent game lke UT2004 is all the difference between a Sempron 2800 with NF4 chipset and a Barton Athlon 3200+ and NF2. Do the math. 50-75 Mhz is needed to ocatch a Sempron with a more modern mobo chipset. Just like I said.....

Now it's been clarified that you meant Sempron vs. Sempron, so the point is now about $$$ ONLY. 939 Sempron versus 754 Sempron. One has a longer upgrade path. That's about the only noticeable difference.

 

[Zap]

Originally posted by: maluckey
Now you can get the nVidia 7800 series for [AGP]...

939 Sempron versus 754 Sempron. One has a longer upgrade path.

The AGP 7800GS is slower and more expensive than the PCIe 7800GT. Time will tell if pricing settles to below the GT, but at this time the extra amount can buy a new motherboard.

You can't just go to your favorite e-tailer such as Newegg and buy a socket 939 Sempron so that's a moot point.

 

[nyker96]

well my sempry 64 2800+ is running about a A64 Venice 2.2 on all benches. At stock runs faster and cooler than my old XP.

 

[carlosd]

Originally posted by: maluckey

Yup!

In 3dMark03 the CPU speed is fairly irrelevant. My best score were at 2100 Mhz. RAM speed and timings, HTT and Sheer GPU power rule that bench.

I was trying to demonstrate that for 3D apps, that you don't need the latest mobo/CPU combo, just the fastest video card around. Now you can get the nVidia 7800 series for NF2 I think I'll stay awhile with my favorite board.

Well semprom 754 is certainly not the latest CPU/mobo combo, and even with the 6800GT the barton @2GHz (running 400MHz higher that the semprom 2800+) is a very limiting factor for latest games (isn' t a 30%+ difference in games like DOOM3 or UT2004 significative enough in favor of a lower clocked semprom?), certainly with a 7800 the baton CPU will be a huge bottleneck as you can see from the bechmarks it is bottlenecking even the 6800GT in a bad way.

Originally posted by: maluckey
carlosd,

You also pointed to the THG sites info on UT2004. It shows 1.6 FPS in a CPU dependent game lke UT2004 is all the difference between a Sempron 2800 with NF4 chipset and a Barton Athlon 3200+ and NF2. Do the math. 50-75 Mhz is needed to ocatch a Sempron with a more modern mobo chipset. Just like I said.....

Well, it seems you forgot barton 3200+ (@2.2GHz) have 600Mhz of advantage over the semprom 2800+@1.6GHz). Just as I said, the Barton have to run 600-700MHz faster to catch up with the semprom 754 . NEVER 50-75MHz!!!, so according to you a barton running at 1.675GHz is going to perform the same than a semprom 754 2800+ @1.6GHz. YES THAT IS WHAT YOU SAY!!!, DO YOU REALLY THINK SO??? so how come the semprom 754 2800+ (1.6GHz) blows out of the water the barton (2GHz) running 400MHz faster?? .

At the same clockspeed the semprom 754 is going to kill barton in EVERYTHING. As long as I remember you were comparing barton and Semprom 754 at the same clockspeed (or what the hell were you comparing??yo gotta have some reference point: same clockspeed?, same prating?) and now you are comparing it to a barton clocked 600Mhz higher. Lets put the semprom @2.2GHz and see how the difference between the semprom 754 and Barton 3200+ is going to grow in gaming very significantly.

 

[maluckey]

Sue me...I Forgot about the stock Mhz difference in the rating system between Sempron and Barton! I should come out of the cave and see whats around sometime. The Barton rating system would have been the Mhz sdifference that I mentioned. Silly me....

Unless you choose to play DOOM3, FEAR or something of that nature, it's not always cost efficient to toss the old setup for a new one is what I am getting at. In games, you can't see the difference between 75 and 190 FPS so most gaming benches are only useful if the absolute minimum framerate is considered. If the absolute mnimun is below 30 FPS, it's an issue if you are serious about gaming.

DivX encoding makes a world of difference, but if it was REALLY important, you'd upgrade to any modern Intel CPU, AMD dual core or an AMD FX 57, not to Sempron.

I myself (and many others) still have headroom (Barton) to run at 210 x 12 (prime95, S&M, Memtest86 stable), One can still run just about ANY game on a trusty 17 inch monitor with average performance and an NF2. You can still spend a bit on a newer vid card as well, though the cost effectiveness tumbles because it can't go with me to a new mobo if I choose to upgrade.

 

[Zap]

Originally posted by: maluckey
DivX encoding makes a world of difference, but if it was REALLY important, you'd upgrade to any modern Intel CPU, AMD dual core or an AMD FX 57, not to Sempron.

It's my understanding that cache and memory speed doesn't have as much to do as pure GHz (platform dependent) in encoding. That is to say that even a Celeron can encode fast if it is high enough GHz.

I know, I know, quoting Tom's Hardware is sinful around here, but those are some easy numbers for me to pull up. That chart is encoding VOB files into XVID - basically ripping a DVD. Going by that chart, different cores make a slight difference, more cache is a bit faster and dual channel is a bit faster. Nothing we don't know. However, the difference is far from night and day.

Let's compare the same GHz with two different CPU/platforms.

core, memory, GHz, cache size, encoding time, OpenGL framerate, cost (right now on Newegg)
Venice, dual channel, 2GHz, 512k cache, 6:38, 170.9, $169
Palermo, single channel, 2GHz, 128k cache, 7:05, 154.8, $111

The difference between the two in encoding? 27 seconds. That's about a 7% difference in encoding performance. Difference in game framerate is 15.1FPS, or about a 10.4% difference. Cost difference? $58, or 52.25%.

How about overclocking you say? Let's use a real-world example. Now, YMMV since your CPUs may not overclock as well or as bad (depends on your POV) as mine, and both our experiences will be different from the third person. My own experience is this... I built my wife's system with a 3200+ Venice. I have a spare Sempron 2600+ sitting around collecting dust. Both happen to max out at about 2.4GHz. Let's do a comparison...

core, memory, GHz, cache size, encoding time, OpenGL framerate, cost (right now on Newegg)
Venice, dual channel, 2.4GHz, 512k cache, 5:40, 193.6, $169
Palermo, single channel, 2.4GHz, 128k cache, 6:04**, 175.4**, $66
**extrapolated by using the 7% and 10.4% numbers we got earlier because AMD doesn't make such a CPU speed.

At my maximum overclock, how much percentage difference in cost for the same 7% difference in encoding performance and an extra 18.2FPS in games (not talking about 10FPS to 28FPS, but 175FPS to 193FPS)? 156% difference.

I know there are a ton of variables that I don't take into account and a different set of CPUs and motherboards may overclock in a completely different manner, but these roughly estimated numbers are interesting to look at nonetheless. I've heard of Venice cores doing 2.5-2.7GHz overclocks, but then again I've also heard of Palermo cores doing as well (though not as common probably because of the low multipliers). Also, my math may be off for percentages... was just called a moron because I couldn't add. :roll:

I guess my whole point for this mental masturbation is that there isn't as much difference in performance as people are hoping/wanting between socket 754 Semprons and socket 939 Venice.

Dual cores are a completely different beast. Single tasking unoptimized software will see no difference from a single core of same GHz while optimized software or extreme multitasking may see tremendous gains - but again at what cost?

 

[ElFenix]

Originally posted by: Zap

Originally posted by: ElFenix
is the OP asking about a barton or a tbred?

The OP didn't mention whether it was a mobile or not. If desktop then a Tbred, mobile a Barton. Also, he edited his post now to say a socket A Sempron 2800+ versus a socket 754 Sempron 2800+. Both are "rated" at 2800+ by AMD. IIRC the socket A Semprons were Tbreds or Thorntons.

I'd still choose the socket 754. Even if AMD says they run close enough to deserve a similar rating, the socket 754 CPU will run cooler, be more durable with the IHS, have a larger HSF choice at this time, have the option of motherboards with PCIe and of course have the option of an AGP motherboard with the retail CPU for $79 at Fry's.

after checking out newegg i'm under the impression that the tbred AXP 2400+ and the tbred sempron 2800+ are the same 2GHz processor. though the 'egg is not listing a 200+ sempron it is easy to see the progression from 2400+ to 2600+ and extrapolate.

 

[Zap]

Basically the same except for a higher FSB.

Man, haven't looked at socket A stuff for a while, seems like the pricing has gone up with the inflated Sempron ratings while the MHz has stayed down.

 

[CP5670]

I used a 6800 GT on a 2.2ghz 512k XP for about six months before switching to a 2.2ghz 512k 939 A64. I basically got major performance boosts in two games (UT2004 and the original Deus Ex) and there was no noticeable difference in any other games or programs, with the exception of Dosbox and Mathematica which ran a bit slower on the 64 for some reason.

 

[carlosd]

Originally posted by: Zap

I guess my whole point for this mental masturbation is that there isn't as much difference in performance as people are hoping/wanting between socket 754 Semprons and socket 939 Venice.

That's exactly what I meant to say.

 

[carlosd]

Originally posted by: maluckey
Sue me...I Forgot about the stock Mhz difference in the rating system between Sempron and Barton! I should come out of the cave and see whats around sometime. The Barton rating system would have been the Mhz sdifference that I mentioned. Silly me....

Unless you choose to play DOOM3, FEAR or something of that nature, it's not always cost efficient to toss the old setup for a new one is what I am getting at. In games, you can't see the difference between 75 and 190 FPS so most gaming benches are only useful if the absolute minimum framerate is considered. If the absolute mnimun is below 30 FPS, it's an issue if you are serious about gaming.

It seems you play very old games. The latests games like Battlefield 2 or Quake 4 are very CPU demanding and sure with an AXP the FPS will cross the 30FPS limit several times not allowing smooth game playing. A small differece in FPSs (in this case it will be no small) can be the difference between good and no so good gameplaying experience.

 

[Fox5]

I'd say a Athlon XP 3200+ or just about any Socket 754 chip is about the bare minimum to have a high end gaming rig nowadays. (there are very few games that an xp 3200+ and most 754 semprons can't manage a 60fps average on)

 

[ElFenix]

tbred 2ghz/333 (xp 2400+/sempron 2800+)
pc mark 04: 3027
pc mark 05: 2358
3dmark2001: 10257
q3: 169 fps

i'll be back with the 754 scores in a bit

 

[maluckey]

The latests games like Battlefield 2 or Quake 4 are very CPU demanding and sure with an AXP the FPS will cross the 30FPS limit several times not allowing smooth game playing. A small differece in FPSs (in this case it will be no small) can be the difference between good and no so good gameplaying experience.

True enough. Below 30 FPS (minimum), gameplay suffers. Running 1024 x 768 or with some setups 1280 x 1024, most older rigs can still play most all games without real issues and without slowdowns. If sheer number of players is any indicator, then most people still (by volume) are playing Half Life2, WarCraft, UT2004 etc. You can play these games with medium details with an OC'ed T-Bred if you really wanted to. You woouldn't get great FPS, but it would be playable.

I also want to clarify for those cunfused (like me) that socket 462 Sempron is beaten by socket 754 Sempron, which is slightly slower than socket 939 Sempron.

 

[ElFenix]

i've just found out that my s754 board doesn't fit in the case i wanted to use it in. (it's an old dell case, i've got the athlon xp running in it, but the front panel connector cable isn't long enough)

so either a) i find a new case
or b) i return it to fry's
now, i suppose i could do the measurements without putting it in a case, but that makes it a bit more difficult to return to fry's (opened and used).

i'll look around at the other cases i have

anyone know how wake/boot on lan works?

 

[ta8689]

Quite honestly, yeah right now it will be a budget machine. But when you want to get a new processor to keep up with your new games or programs or whatever, well i see 754 as hitting the end of the line once the new socket comes out. The fx-60 has been declared as the VERY LAST socket 939 processor to be released. The new one will allow you to use ddr2 i believe. So... 754 is right around the corner from being the next socket A. The extra 100 might keep you from having no option other than to upgrade your mobo, cpu, and vid card if you want to upgrade at all. Yeah...

 

[Calin]

Originally posted by: dug777

Originally posted by: Zap

Originally posted by: dug777
It could be 'ridiculous' i suppose

While i agree with what you say i see no reason to buy s754 when s939 is so cheap now,

Oh teh noes, the spelling nazi. Ya got me there. Diarrhea is something else I have problems with (har har)... Actually I think I have a reasonable grasp of the English language, spelling included, for a second language.

What do you consider "so cheap?" Fry's has a Sempron 2800+ with ECS NFORCE3-A board for $80 (pretty much "guaranteed" overclock to 2GHz). You can also spec out an overclockable PCIe board/CPU deal at Newegg starting at about $125 that has a fighting chance to push the Sempron to 2.4GHz. What was the cheapest socket 939 CPU again?

Pity you didn't bother quoting or answering any of my other points A few extra bucks spent now on a s939 system would serve the OP well down the track (both in terms of performnace and an upgrade path), as well as performing considerably better (especially if you plan to overclock) now

Maybe the original poster wanted to buy now a cheap system, which he would keep a long time. As in my case, I have bought a used Duron 600 when the Athlon 1200 were launched, and still keep the system. I just upgraded the RAM, and nothing more. It is just enough for what I need to do with it, and I didn't want to upgrade it until nowadays. Nothing I could have bought at that time would have an easy upgrade path.