2500+M vs AMD64 CPUs vs P4EE@Tech Report!!!

[DAPUNISHER]

link

It's definitely providing a great price/performance ratio, and if you're getting 2.5-2.7ghz from it with high fsb you are sitting pretty! :beer:

 

[uo7]

Very nice article.

 

[McArra]

Originally posted by: uo7
Very nice article.

 

[slag]

..........And once again, the AMD chip gets stomped in a clear majority of the tests.
Its not even close in the majority of tests. Might as well be testing a cyrix 333 in there if you are worried about price/performance. I'll give you one to play with. Cant beat that price.


I am both an intel and amd owner.

 

[Ronin]

And obviously a leaner towards Intel

 

[uo7]

Originally posted by: slag
..........And once again, the AMD chip gets stomped in a clear majority of the tests.
Its not even close in the majority of tests. Might as well be testing a cyrix 333 in there if you are worried about price/performance. I'll give you one to play with. Cant beat that price.


I am both an intel and amd owner.

How many of those P4's cost 77 bucks ?

 

[slag]

Originally posted by: uo7

Originally posted by: slag
..........And once again, the AMD chip gets stomped in a clear majority of the tests.
Its not even close in the majority of tests. Might as well be testing a cyrix 333 in there if you are worried about price/performance. I'll give you one to play with. Cant beat that price.


I am both an intel and amd owner.

How many of those P4's cost 77 bucks ?

Its all relative.

I seem to recall a certain benchmark that showed the true value of the p4. Benchmarking apps while other apps are running in the background. The p4 with HT absolutely beat the amd to a pulp in those tests. And those are real world. how many of you stop ever app you are working on to play a game? I traded my 2500+ that did 2.3 ghz on air and nf7-s board for a p4 board and 2.6c chip that does 3.27 ghz and the difference is night and day.

Also notice most of those game benchmarks are at 640x480. Right, thats a good test.

Who plays games at 640x480?

Not I.

 

[McArra]

Originally posted by: slag

Originally posted by: uo7

Originally posted by: slag
..........And once again, the AMD chip gets stomped in a clear majority of the tests.
Its not even close in the majority of tests. Might as well be testing a cyrix 333 in there if you are worried about price/performance. I'll give you one to play with. Cant beat that price.


I am both an intel and amd owner.

How many of those P4's cost 77 bucks ?

Its all relative.

I seem to recall a certain benchmark that showed the true value of the p4. Benchmarking apps while other apps are running in the background. The p4 with HT absolutely beat the amd to a pulp in those tests. And those are real world. how many of you stop ever app you are working on to play a game? I traded my 2500+ that did 2.3 ghz on air and nf7-s board for a p4 board and 2.6c chip that does 3.27 ghz and the difference is night and day.

Also notice most of those game benchmarks are at 640x480. Right, thats a good test.

Who plays games at 640x480?

Not I.

Maybe it is in that res to test the CPU and not the video card?

And A64 beats intel in gaming and some other apps.

 

[Ronin]

And what's gonna happen if AMD does Intel one better? What if they come up with oh...perhaps a dual CPU architecture (a TRUE one) on-die? Then HT becomes a simple trick, and AMD has the upper hand again. It's all about progressing technology. These two companies are good for each other, and it only benefits the end user.

You chose to swap, others won't, others will. It's all relative and really a personal preference.

 

[Duvie]

Anyone (Dapunisher) notice how the HT reacted with the P4e chips in the rendering test were they ran 1, 2 and 4 rendering test???? Noticed how it actually got faster and obvious different direction then the northwoods did....

Is this a sign of the improved HT????

I am not surprised by most of the test except I would think a 2.4ghz barton would do a bit better versus the 2.2ghz barton in some of the test....Otherwise it is pretty much what I have seen....

 

[Jeff7181]

slag... you're missing the whole point of that article... I'm not even going to take the time to explain it to you... maybe you should sit back for a minute, forget about what CPU you own and use, forget about what CPU "won" the benchmark... and look at what the article is actually about.

 

[uo7]

I'm not arguing the higher end p4's are a fasterchip. All i'm saying is I don't see the need to spend That kind of money to shave seconds/minutes off my time. If i'm encoding to divx etc i don't sit in front of my comp watching it anyways. As for games i really don't see a need at least in the games i play. Americas army for one. I would rather put the money towards a T.V home/car audio etc. Everyone has there prefrences though.

 

[gwag]

HT is not improved but the new P4's have are less effeciant (despite doubling the L1 and L2 caches) having a crappy 31 stage pipeline, so they will do better in hyperthreading

 

[Duvie]

Originally posted by: gwag
HT is not improved but the new P4's have are less effeciant (despite doubling the L1 and L2 caches) having a crappy 31 stage pipeline, so they will do better in hyperthreading

I agree with that in term the p4e with a single thread and even double thread is always below it same speed northwood however the 4 thread test with the p4e is actually faster then the p4c....Explain that please...

One thing to note as the processor speeds up...ie 2.8e,3.0e, and 3.2e the difference fades away.....


Maybe I am reading it wrong but I assume lower the time the better, right???

 

[Jeff7181]

Originally posted by: gwag
HT is not improved but the new P4's have are less effeciant (despite doubling the L1 and L2 caches) having a crappy 31 stage pipeline, so they will do better in hyperthreading

Not a great way to start off your AT membership...

HT is most definately improved. That's one of the main features of the prescott... improved HT instructions. As people have pointed out in other threads, the fact that a 3.2 Ghz Prescott can perform on par with a 3.2 Ghz Northwood, despite it's slower cache and 50% larger pipeline says A LOT about the Prescott.

*EDIT* As Duvie just pointed out, as clock speed increases, the Prescott gains ground on the Northwood at the same clockspeed. From what I've read, this is mainly due to the larger cache being slower. When the Prescott hits 4 Ghz is when the cache latency is finally equal to the Northwood and there will be no performance penalty as far as cache latency. The 50% larger pipeline should make it easier to scale clockspeed, so 4 Ghz is just around the corner... in fact, people are getting 4 Ghz overclocks already with only minor voltage increases.

 

[hytek369]

good review

 

[DAPUNISHER]

Originally posted by: Duvie
Anyone (Dapunisher) notice how the HT reacted with the P4e chips in the rendering test were they ran 1, 2 and 4 rendering test???? Noticed how it actually got faster and obvious different direction then the northwoods did....

Is this a sign of the improved HT????

I am not surprised by most of the test except I would think a 2.4ghz barton would do a bit better versus the 2.2ghz barton in some of the test....Otherwise it is pretty much what I have seen....

Thanks for pointing that out, I hadn't noticed it. It makes sense that it's the HT improvements Prescott is supposed to have coming into play.

I'd like to add that as far as the performance of the XP-M is concerned the Tech Report is dead wrong for saying the following

We decided against experimenting with bus overclocking, because our intent was to test the processor's capabilities, and bus overclocking would've thrown too many other variables into the equation, such as non-standard PCI and AGP clock speeds. We said it before, but it bears repeating: One of the big advantages of the XP-M is its unlocked multiplier, which allows for substantial performance gains without having to resort to bus overclocking.

The nF2 has PCI/AGP lock so nothing will leave spec, and@440-450DDR dual channel synch mode the scores in anything that thrives on bandwidth would have been better.

 

[Elcs]

Originally posted by: DAPUNISHER


I'd like to add that as far as the performance of the XP-M is concerned the Tech Report is dead wrong for saying the following

We decided against experimenting with bus overclocking, because our intent was to test the processor's capabilities, and bus overclocking would've thrown too many other variables into the equation, such as non-standard PCI and AGP clock speeds. We said it before, but it bears repeating: One of the big advantages of the XP-M is its unlocked multiplier, which allows for substantial performance gains without having to resort to bus overclocking.

The nF2 has PCI/AGP lock so nothing will leave spec, and@440-450DDR dual channel synch mode the scores in anything that thrives on bandwidth would have been better.

Yeah... bit strange. An XP-M with the FSB clocked as high as stable would have been nice to see. My mobile seems happy at 215mhz FSB... no idea whether the chip is holding me back or whether its my TwinMOS PC3200 CH-5... doubt its my NF7-S.

Tech Report are quite alright in my opinion. At least they gave us a rather thorough benching suite....

 

[Jeff7181]

I'm still waiting for someone to donate some PC3700 RAM to me so I can run my Barton at 233x10... my PalmTreePC PC3200 won't overclock an inch =\

 

[Tiorapatea]

Originally posted by: DAPUNISHER
I'd like to add that as far as the performance of the XP-M is concerned the Tech Report is dead wrong for saying the following

We decided against experimenting with bus overclocking, because our intent was to test the processor's capabilities, and bus overclocking would've thrown too many other variables into the equation, such as non-standard PCI and AGP clock speeds. We said it before, but it bears repeating: One of the big advantages of the XP-M is its unlocked multiplier, which allows for substantial performance gains without having to resort to bus overclocking.

The nF2 has PCI/AGP lock so nothing will leave spec, and@440-450DDR dual channel synch mode the scores in anything that thrives on bandwidth would have been better.

That is a very important point, DAPUNISHER.

Noone is saying the XP-M is the ultimate performance processor. The question is, does it provide good price/performance in each of the various applications that the Tech Report benchmarked. A major factor in price/performance is obviously the overrall price of the whole platform - CPU, heatsink and fan (XP-Ms are OEM chips only), motherboard and, importantly, RAM.

I think a good methodology would be to install good but not top quality PC3200 in the test machines and then overclock the FSBs to the maximum stable level, erring on the side of caution. What I would be aiming for is to achieve a level of overclocking that people should be fairly confident of achieving in their home systems without the need to pay the large premium demanded by vendors of RAM rated above PC3200. This premium is especially large if you want to achieve a fast clock and tight timings at the same time.

This would be a more realistic comparison because it is absolutely not a big deal to overclock the front side bus by 10 or 20 MHz using average components. This would help both the P4s and the XP-Ms, although probably not the A64s, since nobody has produced a board for them yet that really does lock the PCI and AGP bus frequencies.

Having said all that, maybe it wouldn't change the rankings very much. I just think it would be a better approach in this type of test.

In the end, TechReport have compared a half-overclocked system with other systems that were not overclocked at all. This is not a great way to investigate price/performance for those prepared to overclock.

Declaration of interest: I own a XP-M.
Apologies: this turned into a bit of a rant, sorry.

 

[slag]

short answer--no, it does not provide a good price/performance. On nearly all tests, the mobile athlon scored in the bottom half of all tests.. How can this be considered anywhere near decent performance???
Big deal, so it sells for $77.00. If it looks like a turd and smells like a turd, guess what! its a turd!

 

[McArra]

Originally posted by: slag
short answer--no, it does not provide a good price/performance. On nearly all tests, the mobile athlon scored in the bottom half of all tests.. How can this be considered anywhere near decent performance???
Big deal, so it sells for $77.00. If it looks like a turd and smells like a turd, guess what! its a turd!

$77 for that performance it's a great price/performance ratio.

 

[adams828]

nice, that's a lot of benchmarks. now if only i had know about the M before i got my 2500+ a while back oh well

 

[Elcs]

Considering a 2500+ can be overclocked to perform what a stock XP 3200+ can do and even exceed it.... Id say its a great price/performance chip.

It seems to me to be suitable for gamer who hasnt got the ££ to upgrade to an AMD 64 system or someone who encodes all day may not see much point in shaving off X number of minutes for an extra wad of ££.

Looking at my recent upgrade:

XP1700+ @ 210x11.5 -> XP-M 2500+ @ 11x215 (not pushed yet)
GeForce4 Ti4400 -> 128mb Radeon 9800 Pro

My graphics card was probably my bottleneck earlier..... Now I have enough power for most games that I'll end up playing for probably 2 years. My GeForce4 Ti4400 is still quite nice but it struggles with some games now so I felt it was time.

PS. My new barton never hits 50C under load and thats with a Thermalright SLK900-A and 92mm Panaflo L1A, ie. nice and quiet.

 

[Megatomic]

Originally posted by: slag
short answer--no, it does not provide a good price/performance. On nearly all tests, the mobile athlon scored in the bottom half of all tests.. How can this be considered anywhere near decent performance???
Big deal, so it sells for $77.00. If it looks like a turd and smells like a turd, guess what! its a turd!

My god, I don't even know where to start with this one...

AXP-M Wolfenstein FPS = 146
P4EE 3.2 FPS = 171.1

Difference = 25.1 fps

AXP-M Cost = $99
P4EE 3.2 Cost = $859

Difference = $760

Cost Difference / FPS Difference = $30.28/fps

So, what in the hell are you going on about slag?