** Now with Poll ** P4 AS FAST AS A64 in DOOM 3 in real-world test

[UzairH]

Ok people some of you, in fact quite a lot of you may have seen this benchmark before, but I haven't seen anyone comment on that here. On every review website when they review the CPU performance in Doom 3 they run the timedemo, with A64 always on top of equivilant P4. But the physics and maybe the AI engines do not run during the timedemo in Doom 3. Xbitlabs did a review some time ago which turned up very interesting results in real-world CPU power in Doom 3:

http://www.xbitlabs.com/articles/cpu/print/doom3-cpu.html

In the initial tests they, like all the other reviews, run the timedemo benchmark which the A64's win. But in real game play runs the P4's actually come out on top. For instance, the P4 3.0 GHz is faster than both A64 3000+ and 3200+. However these Athlons are the older single-channel socket 754 ones. Maybe the newer s939 ones reach parity with the P4, but I don't think the 3% boost from dual-channel the A64 gets will let them cross the P4.

Now tell if you think I am wrong, but I believe this review's results. Interesting stuff eh?

Anyone with access to two similar spec P4 and A64 systems care to verify this conclusion with your tests?

EDIT: My purpose is to prove just how false the assertion by AT and other sites is that for Doom 3 A64 is the best CPU hands down. It seems to be so ingrained that inspite of the link to the article I have posted some people replying here post links to reviews of Doom 3 that run TIMEDEMO, which is not an accurate way of judging gameplay performance in Doom 3.

ANAND PLEASE LOOK INTO THIS. SEE IF YOU CAN RUN SIMILAR OBJECTIVE BENCHES.

 

[UzairH]

Oh yeah, this review has tilted my preference a bit back towards the P4. That, and the fact that I can't find ANY nForce4 mobo here in Pakistan

 

[UzairH]

Could those of you voting also elaborate your reason why, it'll help keep this discussion moving.

 

[MDE]

AMD owns Intel in Doom 3.

Proof

EDIT: You didn't have to post a new thread, just edit your old one.

 

[clarkey01]

"find it remarkable how the A64 3000+ manages to get within 2FPS of the 3.4GHz EE. We have a lowly $160 processor almost besting a $1000 EE. Wow. "


Says it all.

 

[UzairH]

Originally posted by: clarkey01
"find it remarkable how the A64 3000+ manages to get within 2FPS of the 3.4GHz EE. We have a lowly $160 processor almost besting a $1000 EE. Wow. "


Says it all.

Yeah but check the xbitlabs review again... in th timedemo where physics is not calculated the A64's are faster, but in manual, actual gameplay the P4's are ever so slightly faster.

 

[clarkey01]

Originally posted by: UzairH

Originally posted by: clarkey01
"find it remarkable how the A64 3000+ manages to get within 2FPS of the 3.4GHz EE. We have a lowly $160 processor almost besting a $1000 EE. Wow. "


Says it all.

Yeah but check the xbitlabs review again... in th timedemo where physics is not calculated the A64's are faster, but in manual, actual gameplay the P4's are ever so slightly faster.

You may have a point there

 

[Duvie]

Is there potentioal that Doom3 and its physic calculations may be already multithreaded and thus HT wil help it with this area??? If so wont this be a moot point with dual core cpus???

I dont know the answer to the questions!

It seems odd that the time demo, which everyone has been using as a defacto benchmark would not be taken real world use into account in terms of gameplay. I mena the makes of Doom3 know ppl were uses the demp for this purpose. Yo would think they would have clarified this issue if they knew about it. I mean cmon guys. this is the type of crap Intel pulls and gets away with in benchmarks due to being the market lion...not the little guy....


i would say someone like AT needs to also verify this. I dont always trust the methods. wanna be scientist but lack understanding of controlling factors. Also some of this reviews and test are not always done by the main reviewers of the site (Id ont know if this is the case here or not).....


i would put more stock in real world test then synthetic benches, however I though these time demos were as close to real world as it gets. It was a convenient method for ppl to compare...

 

[UzairH]

Originally posted by: Duvie
....
i would put more stock in real world test then synthetic benches, however I though these time demos were as close to real world as it gets. It was a convenient method for ppl to compare...

Exactly. Now when you take Anandtech's Half-Life 2 benchmarks they themselves state that the good thing about HL2 is that during demo playback all the physics etc is also calculated. Nothing of the sort is said or explained in Doom 3 comparison. I believe that the Doom 3 timedemo was designed by id more for benchmarking of GPU, while putting as little stress as possible on the CPU, which might be why no physics code runs during the demo playback.

Bottomline: Doom 3 timedemo makes great GPU bench, but not so good CPU bench.

 

[UzairH]

btw, if true that P4 = A64 (discounting cost difference at value-end: A64 3000+ slightly cheaper than P4 530), then it does throw a big spanner in Anand's and almost everone else's recommendation that A64 is "The" CPU for Doom 3. I don't think anyone from the site is reading this, but do some research into this Anand! Find out how the timedemo is actually different from 'live' gameplay. Then conduct more objective tests. My feeling is that most the newer A64's will be found to be slightly faster vs. P4's than in the xbitlabs article.

 

[bunnyfubbles]

it doesn't matter even if Doom3 is as fast or faster on the P4, all the other games and game benches are lying then? (except for Quake 3 ) Unless you're only concerned about Doom3 that is...

If it is legit it could have something to do with HT, perhaps a test should be done on a dual Opteron system vs. a dual Xeon system...and then see who has the lead.

 

[Avalon]

I think it may have something to do with HT. Xbit should try disabling HT and see if it has any effect.

 

[UzairH]

I am not saying that P4 is better gaming chip, loads of other games where A64 trounces P4. At least Far Cry and Half-Life 2 do. But just discounting a myth really that A64 blows P4 in Doom3.

 

[CheesePoofs]

Oops, I accidentally voted "no I don't care" on the second poll Any way to swich that to the other option?

 

[MDE]

Originally posted by: CheesePoofs
Oops, I accidentally voted "no I don't care" on the second poll Any way to swich that to the other option?

I voted Yes to balance your vote with my eventual I don't care vote.

 

[carlosd]

Who cares if A64 doesn't blow ot the P4 in DOOM??? Here some reasons:

-You are not going to play only DOOM3, or will you? with a 200+ dollars CPU and a 300+ dollars video card??
-A64 blows out the P4 in most the games, DOOM3 is not the only game. Besides I am bored of it!
-At the same performance point AMD is cheaper, especially if you consider the special features that 500s series don't have!.
-A64 chips are much better overclockers, beacuse the performance scales linearly with clock speeds (not the same about the P4)
-A64 are much cooler, getting great overclocks even with stock cooling.
-939 configurations will allow dual core ugrade. Not the same with intel.

Coclusion: The best overall CPU for gaming is crearly and absolutely A64.

 

[MDE]

Don't forget that Quake 4 will be based on a slightly modified Doom 3 engine, so Doom 3 performance does matter.

 

[CheesePoofs]

Don't get so upset ... its just something interesting that I would personally like looked into.

Plus there are other games coming out using the doom 3 engine, such as quake 4.

 

[UzairH]

Originally posted by: carlosd
Who cares if A64 doesn't blow ot the P4 in DOOM??? Here some reasons:

-You are not going to play only DOOM3, or will you? with a 200+ dollars CPU and a 300+ dollars video card??
-A64 blows out the P4 in most the games, DOOM3 is not the only game. Besides I am bored of it!
-At the same performance point AMD is cheaper, especially if you consider the special features that 500s series don't have!.
-A64 chips are much better overclockers, beacuse the performance scales linearly with clock speeds (not the same about the P4)
-A64 are much cooler, getting great overclocks even with stock cooling.
-939 configurations will allow dual core ugrade. Not the same with intel.

Coclusion: The best overall CPU for gaming is crearly and absolutely A64.

Indeddy do A64 is the best for gaming. Even if the two were equal A64's low heat, good overclockability, and 64-bitness make it a better value than the P4.

Problem for me is lack of nForce4 mobos here in Pakistan, my friends are calling me crazy (rightly so) for thinking about uber-expensive (twice actual price) SLI mobo... I like the A64 that much. But knocking off the craziness I'll be forced to go with P4. I just hope a P4 530 will easily oc to 3.4GHz on air - that would provide some consolation.

 

[Avalon]

Originally posted by: UzairH

Originally posted by: carlosd
Who cares if A64 doesn't blow ot the P4 in DOOM??? Here some reasons:

-You are not going to play only DOOM3, or will you? with a 200+ dollars CPU and a 300+ dollars video card??
-A64 blows out the P4 in most the games, DOOM3 is not the only game. Besides I am bored of it!
-At the same performance point AMD is cheaper, especially if you consider the special features that 500s series don't have!.
-A64 chips are much better overclockers, beacuse the performance scales linearly with clock speeds (not the same about the P4)
-A64 are much cooler, getting great overclocks even with stock cooling.
-939 configurations will allow dual core ugrade. Not the same with intel.

Coclusion: The best overall CPU for gaming is crearly and absolutely A64.

Indeddy do A64 is the best for gaming. Even if the two were equal A64's low heat, good overclockability, and 64-bitness make it a better value than the P4.

Problem for me is lack of nForce4 mobos here in Pakistan, my friends are calling me crazy (rightly so) for thinking about uber-expensive (twice actual price) SLI mobo... I like the A64 that much. But knocking off the craziness I'll be forced to go with P4. I just hope a P4 530 will easily oc to 3.4GHz on air - that would provide some consolation.

Just slap some good cooling on it, and you should be good to go. If you've got the video card to back it up, you wouldn't notice the difference if someone swapped in an A64 for your 3.4ghz P4 when you weren't looking. You'll enjoy your new setup, so don't worry.

 

[DrMrLordX]

If it is the physics calculations that keep the P4 competative in "real world" Doom III gameplay, then the introduction of PPUs and/or dual-core CPUs will probably remove any advantage it has in the Doom III engine.

 

[Corporate Thug]

meh, i dont think its wise to select a CPU based on one title, even more so if that one title is DIII

 

[UzairH]

Anyone have any info on whether Doom3 is multithreaded or not? Maybe the physics code is run as a seperate thread, which would account for P4's performance in gameplay vs. timedemo where the physics thread would not be running.

My current rig does not play Doom3 at all well, I have waited for long to play this and others such as HL2 on my soon-to-be rig. So anyone who has this game installed can test the number of threads Doom3 spawns by running ACTUAL gameplay (not just in the menu or a timedemo) and going to task manager and seeing the number of threads the Doom3 process has. Maybe other utilities such as Process Explorer would be useful too.

 

[Acanthus]

The Doom 3 engine *IS* Multithreaded, dual core, hyperthreading, and SMP systems will take advantage of it.

Edit: Proof

Edit2: nv AND intel having an advantage in doom3! The fanbois are taking a double whammy!

 

[Acanthus]

Originally posted by: DrMrLordX
If it is the physics calculations that keep the P4 competative in "real world" Doom III gameplay, then the introduction of PPUs and/or dual-core CPUs will probably remove any advantage it has in the Doom III engine.

The doom III engine doesnt support PPUs, its proprietary technology. (both sony and PhysX)