[H] Cpu Scaling Review

Yea i dont get that ither. The AI should be all CPU. What does the video card performance have to do with AI? The video card just processes the graphics while the CPU carries out the AI and scripts and things of that nature.

What is it about time demo's that make them inaccurate towards real life performance?
 
Heh, Kyle and Brent, i just gotta say you guys are awesome for doing all this work to publish for us readers and putting up with all the crap so many people fling at you for breaking the mold and giving truly BETTER (in every way) evaluations.
 
I think even though everquest2 may be a pay per play game, if its near as taxing as they advertise it would be a very good benchmarking tool if some system of reliable testing could be figured out. whether its pay per play is immaterial, perfomance utilizing the most features is.
 
Excellent article. CPU scaling against the GPU is pretty important, and this one is right from the perspective of a gamer, how a gamer would have the settings and be playing and what to expect.

Not to go off on a tangent, but yeah I really liked the Guildwars alpha (or whatever the E3 thing was) Sort of like UT2004, you can just get in and out of the game quickly and still have fun. PVP is always good, as PVMonster is fun but gets dull fast if thats the only thing you can do. Not having to pay a monthly fee is a bonus (At the end of the year, its $150 I'd rather put to a better videocard)
 
some people have just been mistreated so long that when someone finally treats them well they take it as an insult
 
Oy vey! I'm still very glad I got my X800 XT, I'm just a little worried I will need to move to A64s now instead of my trusty AXP 3200 ... oh darn! :D
Great review too, I trust the Fraps system WAY more than I trust what ATI and nVidia can predict and optimize for. I don't want a card that is going to render the benchmark levels at breakneck speed, I want a card that performs when I play the damn game, and that is what you guys are delivering!

Thanks for this article/review, it helped restore my confidence in the fact that I bough the right card!
 
Considering the review, I think it is good. Most dont actually stress the cards to see what they can really do with alot of action. And thats where you want your card to keep the frames up is in a real action game with lots of people fighting so it dont let you down. With that in mind yes the 800xt seems to be better. Hats off to HardForum for makeing a real world comparison of someone that is serious about playing 3d shooters in multiplayer.
 
well just seeing how nicely the 64s were doing in the review...I have now justified my upcoming purchase of a 939 FX-53
 
I was actually looking forward to this article, but I'm left somewhat disappointed. I'm planning on buying a A64 3000+, and I was hoping to see how it does at stock speed with the different new video cards. Not everyone will be able to OC the 3000+ as high as OCP has, so having the OC benchmarks in there is 'neat' to look at but that doesn't make them truely useful in my case :) I'll also add that I think the benchmark target in terms of resolution and AA/AF should be a min of 30 FPS and an average of 60 FPS. Anyways, keep up the good work guys :)
 
This place is called HARDOCP which I thought meant for overclockers. Why that XP chip was not OC'd leaves me to only one conclusion.

If the OC'd XP chip does as well as a 64 bit AMD them HARDOCP might be shooting themselves in the foot with advertising...banner ads. Buy buy buy the advertisers want and if HArd says a cheap chip can do as well as some 64bit chips then how would Hard be a good place to advertise?
 
JHefile said:
This place is called HARDOCP which I thought meant for overclockers. Why that XP chip was not OC'd leaves me to only one conclusion.

If the OC'd XP chip does as well as a 64 bit AMD them HARDOCP might be shooting themselves in the foot with advertising...banner ads. Buy buy buy the advertisers want and if HArd says a cheap chip can do as well as some 64bit chips then how would Hard be a good place to advertise?
Click to expand...

It might do as well as a low end A64, MAYBE, but i doubt it. An AXP @ 2.4Ghz can barely keep up with a stock 3.0c in most games.
 
I like [H]'s new review system overall. However I would like to see older cards thrown in to show the benefits of the newer cards over the older. Sure we know the 9800's are faster than the 9600's etc, but since many people still use 9600's and 4200 id like to see some comparisons including some of the more common cards as well.

For example, we know the 6800 is faster than the 4200, but how much of an increase in performance is it? And how much of a higher level of detail can you play at?
 
i would have really liked to seen some ut2k4 benchmarks since thats the game that runs they worst on my computer.
 
burningrave101 said:
It might do as well as a low end A64, MAYBE, but i doubt it. An AXP @ 2.4Ghz can barely keep up with a stock 3.0c in most games.
Click to expand...

well I can bet my 2.6AX-m can beat a 3.2ghz intel <not the EE version> SO maybe they SHOULD add in the results

DASHlT
 
burningrave101 said:
It might do as well as a low end A64, MAYBE, but i doubt it. An AXP @ 2.4Ghz can barely keep up with a stock 3.0c in most games.
Click to expand...

Depends on what FSB the AXP-M is at...if it is 240x10 with memory at 2-2-2-11...that 3.0C doesn't stand a chance :)

Guys don't think the oc'd Barton will be as good as an A64...people at XR have had many Barton and A64 rigs. A64 wins...always. It takes a 3.2GHz AXP-M w/ high FSB to match a 2.5-2.6GHz A64.
 
DASHlT said:
well I can bet my 2.6AX-m can beat a 3.2ghz intel <not the EE version> SO maybe they SHOULD add in the results

DASHlT
Click to expand...


but not everyone can get 2.6ghz...sure many can..but other cant...id say keep it middle ground and go for 2.3-2.4ghz
 
DASHlT said:
well I can bet my 2.6AX-m can beat a 3.2ghz intel <not the EE version> SO maybe they SHOULD add in the results

DASHlT
Click to expand...

It probably would beat a 3.2c in alot of gaming benchmarks. Or at least it should considering at 2.6Ghz it should be performing as well as a P4 4Ghz or higher according to how the XP's scale in their ratings per clock rate.
 
DaveX said:
Depends on what FSB the AXP-M is at...if it is 240x10 with memory at 2-2-2-11...that 3.0C doesn't stand a chance :)

Guys don't think the oc'd Barton will be as good as an A64...people at XR have had many Barton and A64 rigs. A64 wins...always. It takes a 3.2GHz AXP-M w/ high FSB to match a 2.5-2.6GHz A64.
Click to expand...

Here are some benchmarks comparing an AXP-M @ 2.4Ghz 200Mhz FSB to quite a few different CPU's.

http://techreport.com/reviews/2004q1/athlonxp-m-2500/index.x?pg=4

It performed on par with a 3.2c in Unreal Tournament 2003.

Lost to a 2.8c in Quake III.

Lost to a 3.0c in Wolfenstein.

Lost to a 3.0c in Commanche 4.

Beat out a 3.4c in Serious Sam.

Scored between a 2.8c and 3.0c in 3dmark03.

AXP @ 2.4Ghz should compare to a Pentium 4 3.6Ghz or in that neighborhood according to AMD's XP rating scale. 3200+ is clocked at 2.2Ghz and suppost to compare to a 3.2c.

An AXP @ 2.6Ghz should compare to a Pentium 4 4Ghz according to that same scale.

100Mhz XP = 200Mhz Intel

Of course it doesn't quite work that way because i havn't seen any XP's able to perform at P4 3.6-4Ghz levels.
 
I would say that it is not a case of whether the A64 wins, it is by how much. I mean to spend twice as much to get (roughly) 10% extra performance doesn't seem great to me.
 
Edgerobber said:
I would say that it is not a case of whether the A64 wins, it is by how much. I mean to spend twice as much to get (roughly) 10% extra performance doesn't seem great to me.
Click to expand...


My thoughts exactly
 
burningrave101 said:
It probably would beat a 3.2c in alot of gaming benchmarks. Or at least it should considering at 2.6Ghz it should be performing as well as a P4 4Ghz or higher according to how the XP's scale in their ratings per clock rate.
Click to expand...

I meant 240x10 for 2.4GHz (who the hell runs 200x12?). And for Bartons not performing as well as oc'd P4's I introduce to you... www.xtremeresources.com

I bet SAE's Mobile Barton can oust a 4GHz Northwood pretty easily...he is using a Mach II GT. Cooling a good Northwood will get you about 4.5GHz...3.2GHz Barton vs. 4.5GHz Northwood...hmmm.

http://service.futuremark.com/compare?2k1=7840841
 
DaveX said:
I meant 240x10 for 2.4GHz (who the hell runs 200x12?). And for Bartons not performing as well as oc'd P4's I introduce to you... www.xtremeresources.com

I bet SAE's Mobile Barton can oust a 4GHz Northwood pretty easily...he is using a Mach II GT. Cooling a good Northwood will get you about 4.5GHz...3.2GHz Barton vs. 4.5GHz Northwood...hmmm.

http://service.futuremark.com/compare?2k1=7840841
Click to expand...


Thanks to Felix88, I spend alot of time over at XR.:)
 
DaveX said:
I meant 240x10 for 2.4GHz (who the hell runs 200x12?). And for Bartons not performing as well as oc'd P4's I introduce to you... www.xtremeresources.com

I bet SAE's Mobile Barton can oust a 4GHz Northwood pretty easily...he is using a Mach II GT. Cooling a good Northwood will get you about 4.5GHz...3.2GHz Barton vs. 4.5GHz Northwood...hmmm.

http://service.futuremark.com/compare?2k1=7840841
Click to expand...

Overclocking the FSB past 200Mhz in a review adds extra variables to the equation. And not everyone can hit 240Mhz FSB. Whats the average for an Athlon XP board? 210-220Mhz FSB?

And people have already had Northwoods over 5Ghz with phase change cooling.

But phase change cooling really isn't what were talking about. Were talking about normal aircooling and maybe at best a little watercooling.

The Barton mobiles usually crap out around 2.5-2.6Ghz or less for most people.

For the price they can't really be beat. But even at those speeds they are still going to get beaten in certain areas by even lower end Pentium 4's.
 
theelviscerator said:
another bench queen.
Click to expand...

ya merlin, saw you over quite a few times!

theelviscerator...I merely showed that link to show everybody what Bartons are capable of reaching...with good aircooling the 35W Mobile 2400+ can hit 2.4-2.5GHz consistently...if you get a 45W Mobile 2600+, it will hit 2.5-2.6GHz with good aircooling...2.6-2.7GHz with watercooling...2.7-2.8GHz with chilled water...and 3.0GHz+ with phase-change. You can beat $87 for a guaranteed 2.4-2.5GHz chip :)
 
burningrave101 said:
Overclocking the FSB past 200Mhz in a review adds extra variables to the equation. And not everyone can hit 240Mhz FSB. Whats the average for an Athlon XP board? 210-220Mhz FSB?
Click to expand...

more like 220-230 FSB on NF7-S, 240-250MHz on DFI Ultra Infinity...users don't hit that high because they don't have good Micron mT or BH-5/BH-6...

burningrave101 said:
And people have already had Northwoods over 5Ghz with phase change cooling.
Click to expand...

not stable...maybe a suicide screenie for LN2...with a Mach II GT the max for a 2.8A Pressie is about 5GHz...those japanese guys hit 5.2GHz with LN2 on a 2.8A for SuperPi benching...

burningrave101 said:
But phase change cooling really isn't what were talking about. Were talking about normal aircooling and maybe at best a little watercooling.

The Barton mobiles usually crap out around 2.5-2.6Ghz or less for most people.

For the price they can't really be beat. But even at those speeds they are still going to get beaten in certain areas by even lower end Pentium 4's.
Click to expand...

Sure a good 3.0C oc'd will beat a normal oc'd Barton anyday...no argument there. But good Northwoods are getting harder to find...its all Prescott these days...those 2.8A chips look really good to me (yes, that's right, me, an AMD fanboy).
 
DaveX said:
ya merlin, saw you over quite a few times!

theelviscerator...I merely showed that link to show everybody what Bartons are capable of reaching...with good aircooling the 35W Mobile 2400+ can hit 2.4-2.5GHz consistently...if you get a 45W Mobile 2600+, it will hit 2.5-2.6GHz with good aircooling...2.6-2.7GHz with watercooling...2.7-2.8GHz with chilled water...and 3.0GHz+ with phase-change. You can beat $87 for a guaranteed 2.4-2.5GHz chip :)
Click to expand...


that was a nice score no doubt he was able to beat my 2001 score by under a 1000 pts...

with several thousand in cooling alone no doubt...
 
theelviscerator said:
that was a nice score no doubt he was able to beat my 2001 score by under a 1000 pts...

with several thousand in cooling alone no doubt...
Click to expand...

if he were to get a X800 Pro...you'd be beat by more than 1k...trust me. then if baker18 lends him a custom gpu block... :)
 
DaveX said:
if he were to get a X800 Pro...you'd be beat by more than 1k...trust me. then if baker18 lends him a custom gpu block... :)
Click to expand...


why are you trying to compare a total custom setup with thousands of dollars in extras against an out of the box setup on air though?

seems I win bang for the buck hands down....
 
theelviscerator said:
his card clocks were above my 800s LOL...
Click to expand...

yes, his 128MB 9800 Pro clocks are not bad for a v-modded card on a 80W TEC...
 
theelviscerator said:
why are you trying to compare a total custom setup with thousands of dollars in extras against an out of the box setup on air though?

seems I win bang for the buck hands down....
Click to expand...

Yea no kidding lol, the 3.0c's are only $200 and most of the new D1 steppings like i have will clock to 3.7-4Ghz on normal aircooling.

In the areas where a CPU really truly matters like multitasking, encodeing, rendering, and heavy applications the P4 would win hands down against a speed demon AXP.

Gaming is a poor CPU comparison IMO because most people with good video cards run high resolution with high AA + AF and the CPU does very little work at all. Most ALL games are video card dependent.

Hell a 3.0c stock would probably perform exactly the same as an FX-53 when you crank the resolution up to 1600x1200 w/ 4x + 8x AF.

I mean seriously, who sits around today playing their games on a high end system at 800x600? Those days are long over.

Thats why i say buying a processor just because its better in gaming at 800x600 or 1024x768 is pretty pointless if your actually wanting a well rounded system.
 
I was thinking about upgrading my 9600 to a new card over this summer but after reading the Cpu scaling article I'm not sure so now because I'm running a XP 2600 o/c to 2.26 on a Nforce 2 board with FSB of 380. Do you guys think it would be a waste of money to put a x800 pro in my system? Thanks
 
Wags317 said:
I was thinking about upgrading my 9600 to a new card over this summer but after reading the Cpu scaling article I'm not sure so now because I'm running a XP 2600 o/c to 2.26 on a Nforce 2 board with FSB of 380. Do you guys think it would be a waste of money to put a x800 pro in my system? Thanks
Click to expand...

No, i think you'd see alot of benefit from going with an X800Pro or 6800GT. Just dont waste your money on an X800XT PE or 6800u for it. Even a 6800nu would be a HUGE improvement and you can get them for $299.
 
theelviscerator said:
you still game with a ti4200?

you are so far out of the game you might as well play solitaire
Click to expand...

I read this again re-reading this thread and about died laughing. Here's why :D ....

I had never played Far Cry until last night. I had put my 6800GT in my main machine and a 9800 Pro in my secondary. ( I had a 9500 Pro in my main and a Ti500 in the other before last night) My buddy gave me a 128 MB 9800 Pro so decided to use that until the other 6800GT gets here......

Far Cry ran like dirt on the 9800 Pro.....I tried to set everything to high....So I'm pretty darn sure you can't run acceptable frame rates on a Ti4200/Ti4400. :p
 
iZero said:
In order to accurately compare two items using quantifiable data (as your graphs representing frame rates do), there needs to be some control measure between the two objects. A timedemo provides repeatable results. Using some person to move through the levels will inherently bring inconsistent results, even if they aren't trying to actively make sure a particular product comes out on top.

Might I suggest creating videos which show your gameplay styles for each set of results. This way there would be no doubt if any sort of bias was seeping into the reviews, as all could examine what's happening.
Click to expand...

video = bandwidth hogging sucker
 
Is there any way you could list the voltages of the OC'd 3000 system? I bought an Abit Kv8 Pro and installed the Beta bios as well, but hit a wall at above 215Mhz FSB. I've increaced the Core voltage, Memory voltage, HTT voltage, running the memory at 333 instead of 400 (It's Corsair TwinX1024-3200C2).

So, the two problems I've found with are:

A. When enabling the Fixed AGP/PCI clock, the Bios will freezes when IDE drives are detected. Whenever I disable them, I can get into XP just fine (running HD on SATA). Did you guys find a way around this?

B. Can't get above FSB 215, doesn't matter whether the memory is 1:1 or underclocked/undertimed, doesn't matter if I increase the voltages of the various components, it just hits that wall and crashes. I have a feeling it's something with the AGP that's causing it, because whenever I get back into XP, it says that the system recovered due from a serious error, but it lists the problem as being with my 9800 Pro card.

Any and all help is appreciated.

BTW- I probably can't give out the exact location of the Beta bios for the KV8 Pro here, but I can tell you that it can be found in the the Abit Socket754 VIA forum.
 
I JUST upgraded from a AXP @ 2400mhz to my A64 2800+ 1800mhz. The 64 wins.
 
burningrave101 said:
Gaming is a poor CPU comparison IMO because most people with good video cards run high resolution with high AA + AF and the CPU does very little work at all. Most ALL games are video card dependent.

Hell a 3.0c stock would probably perform exactly the same as an FX-53 when you crank the resolution up to 1600x1200 w/ 4x + 8x AF.

I mean seriously, who sits around today playing their games on a high end system at 800x600? Those days are long over.

Thats why i say buying a processor just because its better in gaming at 800x600 or 1024x768 is pretty pointless if your actually wanting a well rounded system.
Click to expand...
That's not true.
http://www.ixbt.com/video/itogi-video/0504/itogi-video-u2-wxp-1600.html
http://www.ixbt.com/video/itogi-video/0504/itogi-video-fc-wxp-1600.html
http://www.ixbt.com/video/itogi-video/0504/itogi-video-hl2-wxp-1600.html
http://www.ixbt.com/video/itogi-video/0504/itogi-video-hl2-wxp-aaa-1024.html

According to the benchmarks, the difference is small with last generation video cards (eg. Radeon 9800/9700), but it becomes more noticable (even in the VPU limited setting of 1600x1200..) with the newer video cards (eg. Radeon X800XT PE).

The page with the links to the benchmarks:
http://www.digit-life.com/articles2/gffx/nv40-rx800-4-p4.html

Disdussion about these CPU scaling benchmarks at Beyond3D.
http://www.beyond3d.com/forum/viewtopic.php?t=13631
 
You must log in or register to reply here.
Back
Top