[H] Cpu Scaling Review

This was one of the best X800Pro/Ultra Reviews. I have just one request, Can you guys get a Moblie Barton And run it at 2.4-2.5GHZ. The reason I ask, I want to see if the A64's have a real advantage over the XP's in real life games. And could their be a possiabilty that you might add more games, for example, UT2k4, any builds of next gen games which really stress cpu/video card.
 
I also would have liked to see overclocked mobile barton results in the 2.4-2.6Ghz range. It would have been great to compare against a socket 754 A64 with a paris core like the 2800+ because it's the cheapest.
 
Even I, having an A64, would have liked to have seen a 2.5-2.6ghz Barton to (hopefully) finally discover how much difference there is between the two options in terms of real-world performance.

Other than that, excellent review. It has really become clear as of late that the disparity between real-world performance and methods of measuring that performance has become greater and greater. It's nice to see a hardware site take notice of that disparity and actually try and practically reduce it.
 
Vengance_01 said:
This was one of the best X800Pro/Ultra Reviews. I have just one request, Can you guys get a Moblie Barton And run it at 2.4-2.5GHZ. The reason I ask, I want to see if the A64's have a real advantage over the XP's in real life games. And could their be a possiabilty that you might add more games, for example, UT2k4, any builds of next gen games which really stress cpu/video card.
Click to expand...

Ya i agree I would love to see how a 2.5-2.6ghz mobile barton does...been wanting an A64...but I think my 2.6ghz barton-m is good enough?....perhaps 1 day pleeaaassee :p

DASHlT
 
DocSavage said:
How can you benchmark a MMORPG? Wouldn't you have to have a large group of people stand in the same place with the exact same spells on to ensure accurate repeatable results?
Click to expand...

Actually, it would be quite easy in CoH. There are small instanced missions you could run through with your character. If you repeat the same mission without completing it, you will get the same mobs.

CoH is a good choice because of the whimsical, over-the-top, comic book style. It has more particle effects than any game should have a right to. Basically, each 'power' your character uses is displayed by a different effect. On one of my characters, with all his powers running, there is no less than 6-8 different effects going on at the same time.

http://www.cityofheroes.com/screens/screen_69.jpg

You also gotta love the comic book chat bubbles:

http://www.gamespot.com/pc/rpg/cityofheroes/screens.html?page=175
 
Brent_Justice said:
Our real world gameplay evaluations are manual run-throughs in God mode taking on all the gun firring, all the AI reacting, and the visual damage that occurs to the player.
Click to expand...

just wondering. doesnt AI react different everytime, making different things happen everytime? how did you record the numbers? w/ Fraps?
 
Play Star Wars Galaxies and just stand in a large city at 6pm. Good luck getting good frame rates.

Also, I was a bit disappointed not to see a Barton at ~2.4ghz. Seems like it's one of the most popular CPUs right now.
 
Boooooo! Does anyone who reads the [H] run an axp2500 at stock speeds?
It's probably the most popular OC'ing cpu at this point in time.

If I were to guess, with a decent OC it would nip the heels at the stock a643000, if not best it occasionally.

Until the K8 can make at least 2.5GHz, for $150-175, it just ain't worth it.
 
First off, I wanted to say that I really enjoy the way Hard does their reviews now. Visual time graphs and realworld demos add a lot to the depth of the review. I´ve been around Hard for quite some time and have always enjoyed reading.

With this being said I have a few general comments to make to both the reviewers and the readers:

Comments to people who do not like the new review style

  • Hard is a free information service and complaining about their methods will always be moot. Honestly, be happy they provide you with free hardware information and a place to discuss it. Too many of you are immature. Try using constructive criticism, not angry ranting.
  • You can´t possibly have all of the configurations for graphic settings in a review. You must take what they give you as a base for comparison.
  • Finally, everyone who complains about the new review styles seems to be so fixated on minute details of performance measures. While Hard´s method certainly has problems because of the variables they use in order to achieve similar frame rates, it does show you game performance. As Kyle has stated they are focused on providing the best information for the GAMER. Traditonal review methods no longer accurately demonstrate what it means to be "playable." Take it or leave it there will always be problems with the variables they play with in order to achieve the same FPS, but this is the best review method to convey real world GAMING performance.

Comments to the reviewers
  • The change in review style is awesome!
  • I think that most of your criticism comes from the fact you are tweaking input variables in order to get a common output.
  • Obviously time is a large constraint, but building an interactive exhibit or graph probably would help minimize criticism by readers. So... why not create a small application or JAVA app where the user gets to enter in various inputs (card type, FSAA, resolution, CPU etc) and he can judge the graphs himself. It is difficult to "see the big picture" in the current reviews when the graphs are not on the same axis. This would be one of the most useful review tools because it would allow users complete customization of their needs as well as provide them with the best accurate "gaming" and real world performance.


Let me know what you think. Only trying to help improve. :D
 
Sorry, tried to read the entire thread but many thats alot to keep up with. Well I found the review very good. I like how they test the cards by what I will see and not what companies want me to see. I mean who honestly sits there and watchs the fly-bys? I know I don't, the most I do is run FRAPS in the background while I play...

But for choosing a MMORPG to stress test these cards I would recommend either Final Fantasy XI, Asherons Call 2 or Eve Online. These are the most taxing MMORPGs right now, WoW is NOTHING, Its engine is already dated and City of Heroes is not enough. Star Wars Galaxies has a hard FPS cap of 29FPS (I hit it very often on max settings @1600c1200) so it wouldnt be a good test.

I also would recommend Everquest 2 when released since its possibly the best looking MMORPG to date and SOE has claimed that no PC to date can run it with max settings. It also will be taking advantage of PS3.0 so we can get a nice comparison of it.

If ya got more MMORPG questions just shoot me a PM, I have been playing them since UO and probably have play most if not all retail released MMORPGs...

System Spec:
AMD64 3200
Radeon 9800pro 128MB
1GB DDR400
 
I know that it must have taken a lot of time to do this review, and I found it very informative, but I would also like to say that if you do any more articles like this one in the future, I would appreciate having an overclocked XP mobile on the list to see how it compares to lower-clocked A64 chips.

Also, you are not completely alone in avoiding the normal timedemos in Far Cry because of the AI issue, Maximum PC just loads a savegame and walks straight ahead for a certain amount of time, recording framerates in fraps. I'm sure we would all like Crytek to include a more complete demo system in a patch of some sort.
 
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...

Well, not that I agree with his sentiments, but your comment is bunk. I have a ti4400 and can play every game that has come out up until now (including far cry at high settings) with acceptable framerates (and I'm a FPS junkie). The 4200 is not far off. so get off your high horse just cause you bought into the last generation doesn't mean the ti series isn't a great card and perfectly able to run any game out now with medium to high settings at acceptable framerates.
 
I think many are pissed at the new review style simply because they're focusing on the wrong things. Most reviews simply list a max framerate for a given res and eye candy level. The Geforce does very well in these types of scenarios because it seems able to spike a lot higher in the less complex scenes than the X800's, but conversely it again dips lower in the more complex ones. People are confused - they see the 6800 kicking ass all over, but your reviews show a different real-world picture.

I guess some people just don't like change. The ragged rocks of reality can be a tough place to bash your head open.
 
Bad_Boy said:
just wondering. doesnt AI react different everytime, making different things happen everytime? how did you record the numbers? w/ Fraps?
Click to expand...

Yes, AI does react differently, that is why we run the data multiple times and take the numbers that represent the norm and are repeatable. Yes, Fraps as noted in our review.
 
zlooop said:
First off, I wanted to say that I really enjoy the way Hard does their reviews now. Visual time graphs and realworld demos add a lot to the depth of the review. I´ve been around Hard for quite some time and have always enjoyed reading.

With this being said I have a few general comments to make to both the reviewers and the readers:

Comments to people who do not like the new review style

  • Hard is a free information service and complaining about their methods will always be moot. Honestly, be happy they provide you with free hardware information and a place to discuss it. Too many of you are immature. Try using constructive criticism, not angry ranting.
  • You can´t possibly have all of the configurations for graphic settings in a review. You must take what they give you as a base for comparison.
  • Finally, everyone who complains about the new review styles seems to be so fixated on minute details of performance measures. While Hard´s method certainly has problems because of the variables they use in order to achieve similar frame rates, it does show you game performance. As Kyle has stated they are focused on providing the best information for the GAMER. Traditonal review methods no longer accurately demonstrate what it means to be "playable." Take it or leave it there will always be problems with the variables they play with in order to achieve the same FPS, but this is the best review method to convey real world GAMING performance.

Comments to the reviewers
  • The change in review style is awesome!
  • I think that most of your criticism comes from the fact you are tweaking input variables in order to get a common output.
  • Obviously time is a large constraint, but building an interactive exhibit or graph probably would help minimize criticism by readers. So... why not create a small application or JAVA app where the user gets to enter in various inputs (card type, FSAA, resolution, CPU etc) and he can judge the graphs himself. It is difficult to "see the big picture" in the current reviews when the graphs are not on the same axis. This would be one of the most useful review tools because it would allow users complete customization of their needs as well as provide them with the best accurate "gaming" and real world performance.


Let me know what you think. Only trying to help improve. :D
Click to expand...

Thanks for the support! We have listed to a LOT of feedback and made a lot of changes in the format over the last couple of months, so feedback is far from moot. We try to digest it all and see if the changes or gripes mentioned are valid.
 
Kudos on another great review! You guys just get better every time!

I was very pleased to see the 61.34 drivers get the 6800u on par with the x800pro, hopefully with a few more driver revisions, we may have a real battle on our hands! As for now, it still seems the 800xt PE is the card to have. Now if Ati can just get some good opengl drivers out the door before DOOM3, it looks to be interesting in the next couple months.
 
I realise that setting up all those systems must have been a pain, but could I be another person to chime in asking for the overclocked XP's to be included in the future, especially if you looking at hotly awaited games such as Doom3 and HL2. I for one don't have the cash to get an Athlon 64 system at the moment, so it is important to me to see if fast Athlon XP's can do the business.

One thing about the review, to me it would have been more intuiative to have the results for each GPU is separate plots with all the CPU's (or a selection of) on them, maybe these could be in addition to the plots for each CPU ?
 
First of all, it’s nice to see new tests run with more recent drivers. And second, it’s wonderful to see some different tests for a change. Tests that have useful info for the end user/gamer. I can go to dozens of other sites if I just want to see standard benchmarks run on a single setup and try guess how my results will be or what settings will work best on my system. Plus, call me a cynic, but I really worry about driver updates and how they probably contain more optimizations for the standard benchmarks than they do to actually improve real game performance.
 
Bad_Boy said:
http://firingsquad.com/hardware/x800_pro_athlon_xp/page16.asp
^ is it me or is the cpu's from 2100+ to A64 3400+ no different in 1600x1200 4xaa/8xaf
Click to expand...
Now that review, and not just that page, was very helpful for trying to determine how a 2.6Ghz AXP would perform. If you look at the difference in performance between the 2800+ and the 3200+ you can make an educated guess at the performance at 2.6Ghz. It should be at least 3 times the difference of the 3200+ and the 2800+. That makes a 2.6Ghz AXP as fast as a A64 3400+ in almost all of their tests. Though it would still be nice to not have to guess.
 
oozish said:
Well, not that I agree with his sentiments, but your comment is bunk. I have a ti4400 and can play every game that has come out up until now (including far cry at high settings) with acceptable framerates (and I'm a FPS junkie). The 4200 is not far off. so get off your high horse just cause you bought into the last generation doesn't mean the ti series isn't a great card and perfectly able to run any game out now with medium to high settings at acceptable framerates.
Click to expand...


sorry dude i have seen the screenshots of guys like you and they make far cry look like SoF first version....

and they call acceptable frame rates 19......

you are not even rendering but 60% of the visual info anyway on DX8 card...

I have had the card you are running....3 years ago....so I know full well what it can do...

It runs ok with no AA/AF on at low resolution 1024 and below....

then I got a 9700 pro and was amazed at the diff full AA/AF makes
 
I like the review method, and i trust the numbers. I just wish they didnt aim so low. 30 fps average with dips into the 20's is NOT what i like to play at. I like 60 fps with dips into the 40's. I believe your reviews should aim for that higher mark, not turn everything possible on to get to 30.
 
Brent_Justice said:
We are DEFINITELY going to add that genre of games to our mix.

In your opinion which is the most taxing?

City of Heroes pretty good?
Click to expand...


CoH is an online only game that requires a subscription to even play. That alone I think should preclude it from being used in reviews. However what about the variables that are out of you the tester's control like their servers functioning correctly, clogged pipelines and connections, controlling how many characters, players, spells etc get put into the scene or some monkey boy that decides to jump in the middle of what you are attempting to test around, etc, etc. I feel there would simply be too many factors out of the control of the tester to provide a valuable result with a game that can only be used and run online on their servers. Maybe I've missed it in some of the reviews but have you tested other games and posted their results based on the performance while being played online ?
 
PhyberOptik said:
I think many are pissed at the new review style simply because they're focusing on the wrong things. Most reviews simply list a max framerate for a given res and eye candy level. The Geforce does very well in these types of scenarios because it seems able to spike a lot higher in the less complex scenes than the X800's, but conversely it again dips lower in the more complex ones. People are confused - they see the 6800 kicking ass all over, but your reviews show a different real-world picture.

I guess some people just don't like change. The ragged rocks of reality can be a tough place to bash your head open.
Click to expand...

Other websites use average scores, not maxs. I suggest that if you intend on bashing your head open across every other website out there you actually bother to inform yourself before speaking.
 
Ok i've got a little question that has nothing to do with the GPU performance of ither of these cards. I'm wondering about the performance of the two high-end CPU's, The Pentium 4 3.4c and the Athlon 64 3500+.

If you will look at these two performance results in Far Cry at 1600x1200 w/ 4x AA + 16x AF it would seem to suggest that in reality in "real world" performance, an A64 really isn't any faster then a P4 of the same relative speed.

In fact, the 6800u scored higher on the 3.4c than on the 3500+.

Its the same way in BF:Vietnam between the Pentium 4 3.4c and the Athlon 64 3500+.

Once again, the 6800u on the 3.4c beats the 6800u on the A64 3500+. And this time even the X800XT on the P4 beat the X800XT on the 3500+.

I've been noticeing this more and more in reviews lately. Especially in reviews where they are not meaning to actually compare the CPU's to one another.

I figure the A64 is still better for gaming because of its low latency on-die memory controller. But from the results i'm seeing, there sure isn't a reason to upgrade from a high-end P4 to an A64 unless your just wanting to be ready for 64-bit someday because of the fact the P4 is quite a bit faster in other areas like multitasking, encodeing, rendering, and multithreaded applications.

Is it just because the video cards are doing most all of the work and the CPU's are taken out of the equation totally at that resolution?
 
burningrave101 said:
Ok i've got a little question that has nothing to do with the GPU performance of ither of these cards. I'm wondering about the performance of the two high-end CPU's, The Pentium 4 3.4c and the Athlon 64 3500+.

If you will look at these two performance results in Far Cry at 1600x1200 w/ 4x AA + 16x AF it would seem to suggest that in reality in "real world" performance, an A64 really isn't any faster then a P4 of the same relative speed.

In fact, the 6800u scored higher on the 3.4c than on the 3500+.

Its the same way in BF:Vietnam between the Pentium 4 3.4c and the Athlon 64 3500+.

Once again, the 6800u on the 3.4c beats the 6800u on the A64 3500+. And this time even the X800XT on the P4 beat the X800XT on the 3500+.

I've been noticeing this more and more in reviews lately. Especially in reviews where they are not meaning to actually compare the CPU's to one another.

I figure the A64 is still better for gaming because of its low latency on-die memory controller. But from the results i'm seeing, there sure isn't a reason to upgrade from a high-end P4 to an A64 unless your just wanting to be ready for 64-bit someday because of the fact the P4 is quite a bit faster in other areas like multitasking, encodeing, rendering, and multithreaded applications.
Click to expand...
16x12 4xaa/16xaf and your comparing CPU's :|
 
Cali3350 said:
16x12 4xaa/16xaf and your comparing CPU's :|
Click to expand...

Yes, i'm just curious why there isn't more of a performance difference then that. I know the video cards are doing most all of the work at that resolution but you would think the CPU would play some small role in it. And if the video cards are doing all the work at that resolution, and both settings are the same on both systems, why would the 3.4c perform SLIGHTLY better?

I guess i was just seeing something thats not there lol. I just figured the A64 would have a slight lead instead of the P4. Most likely at 1024x768 the A64 would have a descent lead on the 3.4 P4.

Now i feel stupid for saying anything about it lol :p.
 
At that resolution its almost entirely GPU. If you look at a review where there is focus on the CPU, you'll notice they use low res like 800x600, or 640x480 in order to factor only the CPU's performance. In that case you'll see the A64 shine. Once you get above 1280x1024 you'll see CPU's become almost irrelevant when comparing performance.
 
agar said:
At that resolution its almost entirely GPU. If you look at a review where there is focus on the CPU, you'll notice they use low res like 800x600, or 640x480 in order to factor only the CPU's performance. In that case you'll see the A64 shine. Once you get above 1280x1024 you'll see CPU's become almost irrelevant when comparing performance.
Click to expand...

Yea i knew that, i just figured the A64 would still have at least a 2-3 FPS lead at those settings. And considering alot of us run higher resolutions then 800x600 and 1024x768 with these high-end video cards, it seems like for the most part, the video card is about the only thing thats important in the performance and you should buy your CPU more according to other things you do with the system besides gaming.

But then when you run 3dmark03 at 1024x768 you wont be able to brag about as high of score :D .
 
Good review. I'd like to suggest a chart of the settings for Particle Count, Shadows, Environment and such that you adjusted. Just something simple. Your graphs are easy to read, but in trying to compare say a P4 @ 2.4 to a 2500+ was more difficult because you mixed the specific in game settings with the text.
 
agar said:
At that resolution its almost entirely GPU. If you look at a review where there is focus on the CPU, you'll notice they use low res like 800x600, or 640x480 in order to factor only the CPU's performance. In that case you'll see the A64 shine. Once you get above 1280x1024 you'll see CPU's become almost irrelevant when comparing performance.
Click to expand...
That exactly what I thought, and believed until I saw thses benchmarks.
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 GPU 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
 
Brent_Justice said:
We are DEFINITELY going to add that genre of games to our mix.

In your opinion which is the most taxing?

City of Heroes pretty good?
Click to expand...

SWG is really taxing, and there are tons of graphics quality settings for your enjoyment.
 
Yeah, WoW isn't taxing at all. Heck, it runs fine at 12x10 on my comp and I'm using a total pos! Seriously though, mmorpgs shouldn't be used to test since there are too many variables, they require subscriptions, they have constant game patches that are required to be installed to play (so you can't compare old benches to current ones), etc. There's a reason they're not used to bench hardware.

To the guy who says he prefers min FPS at 40: lower your details... simple, yes?

Amen on the 2500+M at 2.4ghz thingy that everybody (including me) is whining about.
 
If you wish to add MMORPG's to the benchmarking suite, you may wish to consider Guildwars.

www.guildwars.com

Although the minimum requirements that they list for the game are quite...low...if you look at the screenshots, you can see that you'll need a very powerful system to play with full graphics (and yes, those are in game screenshots with no work done to them).

Another good feature about using that MMORPG is that there won't be a monthly subscription fee.
 
Guildwars is not taxing either (ran flawless(no hitch in fps) on max settings for me) and it isnt a MMORPG ;)

It's a diablo 2 clone with a graphical lobby (towns and outposts).
 
|MaguS| said:
Guildwars is not taxing either (ran flawless(no hitch in fps) on max settings for me) and it isnt a MMORPG ;)

It's a diablo 2 clone with a graphical lobby (towns and outposts).
Click to expand...
Well, it is made by Ex-Diablo 2 devs :)
I consider it Diablo 3
 
I guess I'm just one more guy that jumps on the "mobile 2600+ clocked at 2.5-2.8" bus. I just recently purchased a 2600+ mobile because I, like most people I know, don't jump on to the "released within the last 6 months "platform with chipsets that have yet to be revised and tweeked. First generation platforms are usually buggy, exspensive, and therefore not a feasible option for me. Anyway, the mobile XP is still a widely used chip and I would love to see it make its way into some reviews - especially runnging around 2.5-2.8!

Also, no memory tweaking?

Thanks,
Earle
 
Excellent review. Just wanted to add my voice to the numerous cries for at least an estimation of what the mobile@2.4 will do. Is it enough to go up a resolution or will I be slumming at 1024 and perfect frames?
 
Just have to say that I like the way the reviews have been done . The defense that there are plenty of cookie cutter reviews that do all "apples to apples" is valid. Also including the min/max/avg is great.
I am curious why there is such a difference in most of the other reviews compared to the HOCP though. Most places are showing a much closer race.
Now about the timedemo vs the fraps thing. If a timedemo doesnt include AI then it is almost pure graphics right? It makes sense that the AI could slow down the benchmarks. But why would that change the realitive speed of the cards? Wouldnt the AI be more of CPU limiting situation? Why would the AI slow down one card more than the other?
Somewhere along the way the numbers do not seem right. Im not pointing the finger either way but there certainly is a difference here. Just wondering what is causing it.
 
You must log in or register to reply here.
Back
Top