5870 (vs GTX 280) Results: Confusing

XFX 5870 arrived this morning, and I wasted no time in putting it to the test. The results? Not what I expected, and confusing after seeing how this thing supposedly performs in every review.

It runs great... except for the fact that it's actually friggin' slower than my GTX 280, and sometimes by a rather large margin.

The following results were played exactly the same (weapons, explosions etc.) in exactly the same spots. I made sure to be dead-on with everything I did and where.

Note: multiple benches were done; these numbers are a general average of the results I've been getting for the last three hours, but I've gotten even worse than what I'm listing.

Rig: 5870 (stock), E8600 (stock), ASUS Rampage Forumula, 2GB RAM, XP 32, 1920x1200 res
Driver: (disc)

Left 4 Dead (No Mercy, 1st map, horde)
8xAA/16xAF; settings maxed

GTX 280
min: 39
max: 63
avg: 57.7

5870
min: 34
max: 62
avg: 50.7

Min: 5+ FPS slower
Avg: 7+ FPS slower

Fallout 3 (same fight/area; same weapons/explosions)
8xAA/16xAF, Ultra, settings maxed

GTX 280
min: 35
max: 62
avg: 54.7

5870
min: 28
max: 62
avg: 45.8

Min: 7 FPS slower
Avg: 10 FPS slower

FurMark:
Max Temp: 50C
FPS: min=27 max=51 avg=34

What's the deal here with all these numbers? These are not event the worst of the differences. Through benching multiple times, there were even bigger gaps between GPU's, upwards of 15 FPS.

The 5870 supposedly "trounced" the GTX 285 in every review done...?
It's supposed to be on-par, relatively, with a GTX 295?

There's nothing wrong with my system or how I'm running, it's just running slower at the same settings.

Only exception is Crysis: Warhead (Enthusiast, 2xAA/16xAF) which generally seemed to actually run faster/better/smoother; faster by about 4-10fps mix/max/avg (that was impressive) in some places, but relatively the same otherwise, with the exception of feeling a bit "smoother". So, that impressed me, at times.

Fan set to 60% for all games, only "loud" in terms of air, but actually very quite. Very surprising... and the temps and sound are the only advantage I'm seeing with the 5870, which makes no sense what-so-ever.

I did multiple benchmarks, and got some results that were even worse, with the FPS being even further apart, upwards of 10+ less in mix/max/avg in all games I've tested thus far.

Temps: excellent... astounding, really. 30C idle w/ fan @ 30%. Never hit over 48C with fan @ 60% under load! Definitely impressive.
Visuals: no noticable difference in anything, which was surprising, given ATi's usual color and AA/AF advantage over nVidia that I've always noticed over the years.
Performance: How the hell did this beat-out the GTX 285 in all the reviews done?

While sometimes hard to tell with the eye, sometimes the difference was visually noticable; numbers show the difference difference.

It's not my system... unless my system just happens to "like" nVidia better

Reinstalled my GTX 280, and results were back to normal, beating the 5870 at stock settings, sometimes by a large margin.

Disc driver, updated XP driver...
a) the driver on disc does run fine; b) no more than a max of 5FPS (and that's pushing it) will ever been seen with an updated driver, and that usually only happens in specific games.

I've pushed this thing all morning, benched countless times in the same games, same settings and the same exact methods.

Not here to bitch... here to legitimately figure out what the deal is with this thing, considering I'm getting results that are the polar opposite of every review out there. I was looking forward to perhaps moving over to/sticking with ATi's newest, but with the results I'm getting... genuinely looking for help/feedback here.

Might be hitting some sort of driver bloat or issue with overhead. You might want to push your CPU a bit to see if that improves rates with the tests or games.

I don't think he is CPU limited at 1920x1200 with 8xAA on. Have you tried monitoring the clocks under load? My friend's GTX275 would downclock itself even when temps were fine, took a bios update to resolve the issue.

+1

Driver bloat, and all. I reinstall my system everytime I get a new replacement card (sound, ethernet, graphics, etc). New, I just install the drivers. Replacement, I wanna clean start.

Also, if you have a nforce chipset... (I'm just joking, but it may be a tad more optimised for nvidia cards, normal for all others?)

Also, see if your card is running at full PCIeX16. I got a card that, for some unknown reason, tripped the PCIeX1 setting, and it was slow. Open up GPUz, and woah! (the setting should be in the BIOS).

A few things... Windows XP x86, 2GB ram, stock CPU....

I have a feeling these things are just bottlenecking the ATI card. I've noticed a lot of the performance increase of the benchmarks are in DX10. Which you running XP will never even see. When you look at DX9 performance of the HD5870 vs a GTX theres really not a whole lot to gain.

Make sure the PCI-e slot is running 16x and not 1x or something else. Make sure to uninstall Nvidia drivers, run driver sweeper then install the ATI Catalyst drivers, be sure to use the hotfix ones on the ATI site as they are 9.10 betas. And for God's sake - overclock that CPU. I had a few 8500s that did 4.3-4.5 easy. You should at least be able to push 3.6-3.8 even on air.

Yeah I hate to say since I love XP, but I have a feeling this 5870 is built around W7 and totally optimized for that. I think this card just shows how dated Xp really is. All the driver enhancements are for W7 I'm sure...

Truly appreciate the responses.

In response...

Despite running XP x86, with a GPU this powerful and everything else I'm running, I really cant see how it would affect performance of the 5870 like this. You could be right, but I would be shocked to see something like that, given the power of the 5870, even if I'm "only" running 2GB RAM (more on that later).

- Running @ PCI-E 2.0 x16 (according to GPU-Z, so good there)

- CPU definitely not the issue

- Driver bloat... eh... never seen driver bloat kill a GPU's performance like that, but I do know what you mean. However, I clean everything out from the registry with both DC Pro and manually (I've been doing this a long time).

- The 5870 is not down-clocking while in-game. I alt+tab in and out to check temps and settings, and everything is running as it should in terms of the GPU's settings.

Blasted through, reinstalled both again, cleaned out nice... same friggin' results.

FurMark - GTX 280
FPS min=35 max=54 avg=41 ...trashes the 5870 by 7 FPS min across the board every time.

L4D - again, same settings... min=37 max=62 avg=52.2 - again, beats out the 5870 (and sometimes by more than this; I just benched it three times)

Fallout 3 - again, same settings... min=24 max=60 avg=42 - this bench showing a little lower than the 5870 on Ultra, but the second and third benches I did were higher on the GTX 280 by a bit.

Crysis: Warhed - again, same settings... pretty much evened-out; slight performance increase in places, but mostly less and/or equal in others.

Each game just tested again between both, and benched in the same places three times each, same situations, same settings and same gameplay.

Since the 5870 beat-out the GTX 285 in every other review/test by every site out there, this situation I cannot explain...

...and it's really frustrating the hell outta me.

Just so you guys are aware, I'm definitely not a novice. I don't know "everything", but I've been building systems for the last 9 years (learned a lot from here, actually), so I can tell you I'm definitely not missing anything that I've ever seen happen before.

P.S.
Ha ha... so intense on the CPU OC'ing .

I do not OC anymore. The E8600 already runs stock @ 3.33, so moving it up to even 3.8 is not going to make a huge difference.

Running XP x86 I honestly do not feel is going to "bottleneck" the GPU like this (I've honestly never seen that happen with any GPU), nor is 2GB RAM. If I were under 2GB RAM, sure... but then, I'd be getting poor performance no matter what GPU I was running. Sure thing, more RAM could not hurt, but not many games take huge advantage of it (I've done a lot of direct comparisons myself with mine and other systems) and 2GB RAM would not be killing the performance of a GPU that should be crushing the GTX 280.

Either way, as I've said: running in PCI-E 16, all drivers cleaned out thoroughly always, all settings checked and re-checked and watched very carefully. On my mother's eyes, I'm telling ya... just don't get it...

Honestly, even running in DX9, the 5870 should be the stronger performer due to the hardware.

Even if I was slightly bottlenecked by something. If the 5870 trashed the GTX 285 in [H]'s (and everyone else) tests, than there should be a difference here... for the better, it should not be running slower, given that it's more powerful (or supposed to be). The performance I get from the GTX 280 is hefty with everything maxed.

Not "arguing" at all with the feedback here at all, just not clear here because the OS and/or the RAM should definitely not be an issue to that much of an affect, especially given my currently performance.

I'm just responding to what/how I'm running and what my experience has been and what it is now... and I'm completely at a loss to explain this situation.

Not sure if I should return the GPU and stick with the GTX 280, if there's just something "unavoidable" here, but that cant be possible... could it? How could the OS, or even 2GB RAM actually bottleneck this GPU like this? It's not like 2GB of high-end RAM is considered "low-end", even these days.

GTX 280 is a friggin' beast, and it runs like one on my system! Heh.

Man, frustrating...

But, what ever is happening for me here, maybe can be explained directly at some point, and if it seriously is something with the OS (which I just cant imagine, not to this degree, to bring it down from being equal to a 295 to less than a 280?), then maybe seeing my situation will help others... if it can even really be answered with something concrete.

You could try 3dmark Vantage to see if there's something wrong with your results.

astrallite said:

You could try 3dmark Vantage to see if there's something wrong with your results.

Click to expand...

I actually was going to earlier, but it's only supported by Vista.

Is there a version for XP? I'm not finding one...

This is just speculation, but my money would be on this being related to XP64 drivers.

If you really wanted to test it out, you could Win 7 64 RC to see if that helps.

Edit: oops, nevermind, misread your other posts. XP shouldn't be bottle-necking like that in general. Very odd.

My bet would be the driver bloat, it seems even if hes bottlenecking with the new card he'd be getting worse results with the GTX280

I think you should move up to Vista/7 64bit and get more ram, then see if the results are still the same.

You're obviously running into something, probably OS related, since they're both coming up pretty much even. Maybe someone else here can do a test on 32bit XP or something and see if it's reproducible.

verylostindeed said:

This is just speculation, but my money would be on this being related to XP64 drivers.

If you really wanted to test it out, you could Win 7 64 RC to see if that helps.

Click to expand...

XP 64 drivers? Meaning...?

I'm running XP32.

"Win 7 64 RC"? Not sure what you mean, but I don't have Win 7. I also don't run 64-bit anything, due to the fact that I use a lot of important proggies I use that don't have/do well with 64-bit drivers.

Truthfully, I cannot see how the OS is the problem. This GPU was being developed simultaneously, so I don't see how it's "optimized" for Win 7... certainly not to a point where something this powerful would be hindered to perform below a GTX 280.

That's a pretty big problem... one I've never seen happen with any OS/GPU config in any gen of GPU's.

But, who knows... it's certainly odd, and more than disappointing. Guess I'm sticking with the GTX 280 and possibly just nVidia in general if the architecture of ATi's new GPU's is such that the OS could actually be an issue that would hinder it's hardware this much.

I find that... crazy...

Just an idea. I can't say with a certainty that your xp/2gb build is the issue or not, but your results are the outlier from many, many respected sources. I am thinking it's something in your build that is the issue, in fact I would bet on it. Which I could offer you some other explanation other than your "base" system is long in the tooth (XP was great, it's now 2 generations old. And 2 gb of mem is what comes stock on emachines these days)m but maybe it's just time to move on. What I would be really interested in is if you can get the same level of performance with lesser Gfx card. Pop in a 9800gt or some such and re run the tests, if your able that is. I am guessing you have some type of I/O bottleneck that is resulting in a "perfect storm" of suck and frustration for you. And I concede I have never heard of anything like this before really, save for being CPU bottlenecked, Then again the 58XX is a new piece of hardware so I am sure issues will come up.

Abit667 said:

I think you should move up to Vista/7 64bit and get more ram, then see if the results are still the same.

You're obviously running into something, probably OS related, since they're both coming up pretty much even. Maybe someone else here can do a test on 32bit XP or something and see if it's reproducible.

Click to expand...

That would be a great thing, if someone else could do that.

You could be right, as you said, since they're both coming up relatively even in performance, but the GTX 280 is beating it out.

Again, I'd find that to be so friggin' odd, though... never seen that happen before with any GPU.

Also, I mentioned above, I don't run 64-bit due to some of the progs I use that don't work well in the environment. I could try it, I suppose. I'd have to snag a copy of Win 7 64, but man... so much to deal over something I find so... odd.

I don't think more RAM is necessarily going to help either. 2GB of solid RAM should not hold back this GPU like this.

Eh, so maybe it is an OS thing.

Something AMD needs to know about, if they don't already! Nothing ever mentioned regarding OS performance until now, lol.

Slachtbeest said:

XP 64 drivers? Meaning...?

I'm running XP32.

"Win 7 64 RC"? Not sure what you mean, but I don't have Win 7. I also don't run 64-bit anything, due to the fact that I use a lot of important proggies I use that don't have/do well with 64-bit drivers.

Truthfully, I cannot see how the OS is the problem. This GPU was being developed simultaneously, so I don't see how it's "optimized" for Win 7... certainly not to a point where something this powerful would be hindered to perform below a GTX 280.

That's a pretty big problem... one I've never seen happen with any OS/GPU config in any gen of GPU's.

But, who knows... it's certainly odd, and more than disappointing. Guess I'm sticking with the GTX 280 and possibly just nVidia in general if the architecture of ATi's new GPU's is such that the OS could actually be an issue that would hinder it's hardware this much.

I find that... crazy...

Click to expand...

I edited the prior post; for some reason thought you were running XP 64.

You can download Windows 7 64 legitimately and for free. You might want to just try it out with a dual-boot to see if that helps at all. XP should not be causing this problem, but the OS is very likely the culprit here in some combination with ATI's (god awful) current drivers for XP.

That's the cheapest thing to try in any event.

Ghettobox said:

Just an idea. I can't say with a certainty that your xp/2gb build is the issue or not, but your results are the outlier from many, many respected sources. I am thinking it's something in your build that is the issue, in fact I would bet on it. Which I could offer you some other explanation other than your "base" system is long in the tooth (XP was great, it's now 2 generations old. And 2 gb of mem is what comes stock on emachines these days)m but maybe it's just time to move on. What I would be really interested in is if you can get the same level of performance with lesser Gfx card. Pop in a 9800gt or some such and re run the tests, if your able that is. I am guessing you have some type of I/O bottleneck that is resulting in a "perfect storm" of suck and frustration for you. And I concede I have never heard of anything like this before really, save for being CPU bottlenecked, Then again the 58XX is a new piece of hardware so I am sure issues will come up.

Click to expand...

If by "base system" you're referring strictly to XP32 and 2GB of RAM... you (all of you) could be possibly be right, I suppose.

Again, I find that really odd, as you've said, I've never seen anything like this happen before.

As for the rest of my hardware, definitely not long in the tooth.

Like I said, I could snag a copy of Win 7, even 64-bit, and try it again. Then add more RAM, and test it some more.

I don't have any older GPU's anymore that I could test with... well, there's the 8800 GTX in my wife's system, but that thing is still a beast and ran great on this system at this res, so not sure that would be worth it.

what is your setting for the furmark bench?

my 5870 at 1920x1200 MSAA 0X is about avg 117

bangmal said:

what is your setting for the furmark bench?

my 5870 at 1920x1200 MSAA 0X is about avg 117

Click to expand...

I posted those results above... 8xAA for both GPUs (just check the posts above ya).

verylostindeed said:

I edited the prior post; for some reason thought you were running XP 64.

You can download Windows 7 64 legitimately and for free. You might want to just try it out with a dual-boot to see if that helps at all. XP should not be causing this problem, but the OS is very likely the culprit here in some combination with ATI's (god awful) current drivers for XP.

That's the cheapest thing to try in any event.

Click to expand...

No prob... yeah, I'm running 32-bit.

Hmmm... perhaps I will try your suggestions. Did not realize you could dl Win 7 legit... for free? Full version? Or upgrade?

But, as you've said yourself, XP should not be causing this problem, yet at the same time, maybe the idea of the OS mixed with the current drivers... anything is possible, I suppose, no matter how odd... especially with hardware! LOL.

Crazy!

unless you want to believe that every single published review and benchmark are complete bullshit, obviously the problem is on your end, not the card. try turning off all aa and af on both cards, and run the same benchmarks. if the results are more in line with what is out there about the card, add aa and af back in one at a time and see if its a prob with one of those. anyway, i think a clean install would prob clear this up.

where would he be able to get a copy of win7 for free? i thought the program was over, no more keys or dls.

Well I'm leaning towards either 2GB of ram just not being enough and its just bottlenecking or the CPU still. Think of these two powerful cards and neither one is able to really stretch their legs - so you results are just the result of the bottleneck. Maybe the GTX is just dealing with it better which is why your results are really within a margin of error of each other. Something is keeping the HD5870 from moving any faster - what is it? As you've stated and gone through it's probably not the OS or the drivers - which leaves hardware. I would suspect your HD5870 is sitting there waiting for information from whatever is the bottleneck. 2GB of isnt enough anymore. Alot of games can load maps that are almost 2GB add to that the OS layer and you can easily fill 2GB ram. If the machine is bottlenecked there and just cant load anymore information then that may be why the cards are evened up. If the CPU just isnt able to keep the data coming the same thing is gonna happen. So from what you given us, it doesnt sound like it's a software / OS issue unless the Catalyst in XP x86 are just bugged. Its now a PSU issue as we all know the HD5870 isnt even using as much power as the GTX.

I'd try to OC the cpu, re-benchmark - and see what happens. Since thats pretty quick, easy and free its your best first shot in solving this. Also, you can get a beta release of Win7 legit and free and see how that runs too.

The E8600 could easily be holding back cards of the GTX280 and 5870s caliber.
You will get higher framerates with both cards at 1920 res if you clock the CPU.
This may not be the issue you are having, but if the 5870 needs more CPU than 280....

Every new generation of gfx card requires more CPU so I believe clocking the CPU will help.

Their's always one person that gets the short end of the stick And looks like the 5870 picked you sorry to say...

Slachtbeest said:

If by "base system" you're referring strictly to XP32 and 2GB of RAM... you (all of you) could be possibly be right, I suppose.

Again, I find that really odd, as you've said, I've never seen anything like this happen before.

As for the rest of my hardware, definitely not long in the tooth.

Like I said, I could snag a copy of Win 7, even 64-bit, and try it again. Then add more RAM, and test it some more.

I don't have any older GPU's anymore that I could test with... well, there's the 8800 GTX in my wife's system, but that thing is still a beast and ran great on this system at this res, so not sure that would be worth it.

Click to expand...

By "base" system that was all I meant. Sorry I couldn't find a better term . I'm just thinking of all the reviews, all of them are running i7's with 6+GB of ram, and win 7. They were all seeing moderate to impressive gains. I don't doubt your CPU is up to the task, and likely the rest of your build is also, or at the least I have no reason to doubt your statement that it is. So when we look at the variables, we come down to only a couple of things that stand out. XP, and the 2gb of memory. I am not going to try to sell you on win7, but with the XP vm any app you have should run fine (IF your CPU supports Virtualization, consult Intel site for compatibility), and I am guessing you will see better performance. I feel your going to be butting your head against the wall, at the least I can assume the target market for the 5870 is going to leave you in the minority with that configuration. Making troubleshooting all the more difficult.

vanilla_guerilla said:

unless you want to believe that every single published review and benchmark are complete bullshit, obviously the problem is on your end, not the card. try turning off all aa and af on both cards, and run the same benchmarks. if the results are more in line with what is out there about the card, add aa and af back in one at a time and see if its a prob with one of those. anyway, i think a clean install would prob clear this up.

where would he be able to get a copy of win7 for free? i thought the program was over, no more keys or dls.

Click to expand...

No, of course I do not believe all the test results are "bullshit", but it's not necessarily on "my end". No one has tested this on XP x86. Could be a problem with this GPU specifically.

Who knows... if it is on my end due to the OS, that's something AMD should have been aware of from the beginning and made public.

I don't see how there would be a specific "problem" with AA or AF individually, but I can try it.

I could also try a clean install, but with what OS? I have a "back-up" drive that I keep a fresh install of XP on that has nothing else on it but the OS itself. However, even this drive is pretty damn clean (just did a new install recently).

But, if people think XP is the "problem" (again, which I can't imagine, nor some others here are necessarily agreeing with) then I'm stuck unless I can get a hold of Win 7.

i dont think xp is necessarily the prob, or 32 bit. i think that many, many pc's are still configged this way, and no way could amd have ignored this. its just that many times when i have upgraded to a new vid card, i got crappy results until i did a fresh install. try the xp image that you have.

This is making a mountain out of a molehill.
Try clocking your CPU.

Ghettobox said:

By "base" system that was all I meant. Sorry I couldn't find a better term . I'm just thinking of all the reviews, all of them are running i7's with 6+GB of ram, and win 7. They were all seeing moderate to impressive gains. I don't doubt your CPU is up to the task, and likely the rest of your build is also, or at the least I have no reason to doubt your statement that it is. So when we look at the variables, we come down to only a couple of things that stand out. XP, and the 2gb of memory. I am not going to try to sell you on win7, but with the XP vm any app you have should run fine (IF your CPU supports Virtualization, consult Intel site for compatibility), and I am guessing you will see better performance. I feel your going to be butting your head against the wall, at the least I can assume the target market for the 5870 is going to leave you in the minority with that configuration. Making troubleshooting all the more difficult.

Click to expand...

Again, appreciate the in-depth response.

Well, you'd be right to trust what I'm telling you regarding my results. I'd not just "make something up" .

Also, it's true that it can only come down to a few things here... the variables being either XP and/or the RAM, but perhaps the GPU itself (though I don't feel so).

But again, I'd think AMD would have tested these things out, and let let people/the industry know, especially since so many people are still running XP. You'd think they'd have tested with different specs regarding RAM etc. and said, "Hey! 2GB of RAM just ain't gonna do it with this GPU and XP!", ha ha.

I'd have no probs moving to Win 7, as long as I can get everything to run right. Matter of fact, been planning on it.

So, I suppose the question is: do I hold onto this GPU until I can get Win 7, drop in more RAM, then see?

Because, as you've said, every review done has been with Win 7.

if your machine comes up to the minimum requirements to run this card, dont bother overclocking. if it was bottle necking the 5870 a lot more than the 280, your results would come out evenly, not skewed in favor of the 280. and i think enough guys are still using xp (many of the benchmarking guys at xs still use xp performance edition) that if it was a problem, it would have gotten out, or a warning on the box. re install xp.

Nenu said:

This is making a mountain out of a molehill.
Try clocking your CPU.

Click to expand...

Uh... no offense, but it's not making a mountain out of a molehill.
I'm running the highest-end dual-core on the market, and @ 3.33, that's not the problem.

vanilla_guerilla said:

i dont think xp is necessarily the prob, or 32 bit. i think that many, many pc's are still configged this way, and no way could amd have ignored this. its just that many times when i have upgraded to a new vid card, i got crappy results until i did a fresh install. try the xp image that you have.

Click to expand...

I'm starting to think this more and more... if you (and some others) are not feeling that XP or 32-bit could really be doing this, and I don't think AMD just "skipped over" this either (I was being facetious), then maybe I will do that very thing... slap it in on the other drive with the fresh XP install and see what happens.

vanilla_guerilla said:

if your machine comes up to the minimum requirements to run this card, dont bother overclocking. if it was bottle necking the 5870 a lot more than the 280, your results would come out evenly, not skewed in favor of the 280. and i think enough guys are still using xp (many of the benchmarking guys at xs still use xp performance edition) that if it was a problem, it would have gotten out, or a warning on the box. re install xp.

Click to expand...

Exactly.... and they are relatively even, but the 5870 is definitely falling behind a bit... so, I guess you're just backing-up what you were saying about a fresh install here?

yes. and 2gb ram is plenty of ram for benchmarking. you have 1gb ram on the card.

Slachtbeest said:

Uh... no offense, but it's not making a mountain out of a molehill.
I'm running the highest-end dual-core on the market, and @ 3.33, that's not the problem

Click to expand...

I'm running a GTX260 on an E8400 at 4.1GHZ on XP32.
Clocking the CPU makes a big difference.

pls read my first post too, you seem to have missed it.

Honestly i would do a windows reinstall.. Those numbers are WAY off...
I know i have had driver bloat cause really bad performance numbers before but this was when i went from nvidia to ati back from the ti4200 to ati- 9800 years ago and this was in windows XP.... vista/7 handle these things alot better so i would reinstall windows. I havent had any of those issues going from ati back to nvidia but i have done so on new os's 1900xtx to 8800gt and now i will be going back to ati again....

oh also.... do you have vsync turned on or forced?

vanilla_guerilla said:

yes. and 2gb ram is plenty of ram for benchmarking. you have 1gb ram on the card.

Click to expand...

Yeah for benchmarking it is. Then again I am going to guess he's not running some gutted version of XP, probably has some services/prog's running in the background etc. I think we are looking at real world performance, on a real world machine. With 2gb of ram, and 1gb on the card he is at 3gb of a theoretical 3.5gb limit for a 32bit OS. I have no idea what kind of optimizing he has done, or hasn't done but the machine is pretty much at the limit for xp32 bit. I still would contend most gamer's have moved to Vista/7 and it's going to make his problem all the harder to track down. And generally minimum spec's are just that when it come's to games/hardware. We all know the more you can exceed them the better the real world experience will be. I am not saying it's the issue (xp, or 2gb of ram) but there is a mountain of evidence that point's that way, I'm really not sure how any other stance can be defended.

vanilla_guerilla said:

yes. and 2gb ram is plenty of ram for benchmarking. you have 1gb ram on the card.

Click to expand...

OK, so this is at least one vote, from your responses, going for clean XP install could solve the problem.

Nenu said:

I'm running a GTX260 on an E8400 at 4.1GHZ on XP32.
Clocking the CPU makes a big difference.

pls read my first post too, you seem to have missed it.

Click to expand...

Sorry, I think I did miss it somehow.

Certainly, OC'ing can definitely have a big impact, depending on how far you push it, but as said, I'm not really into OC'ing things any more. Over the years, I sort went into "any high-end hardware should do find on it's own without having to be OC'd" mentality, and just stopped with all the bother of doing it.

Plus, I really don't think that, regardless of how powerful these GPU's are, that OC'ing the CPU is what's necessary to get the 5870 to run the way it should. There has got to be another issue, because the E8600, while no longer the newest CPU, is definitely (even at stock) the fastest non-extreme dual-core on the market and still holds it's own even at stock speeds.

I'll move to Win 7 upon it's release, and definitely be bumping up to at least 4GB RAM, but as much as I appreciate the suggestion, I'm not into the OC'ing thing anymore.

So, I'm not sure what to do... hold onto the GPU, try a fresh install, if that doesn't work then keep it and wait for Win 7?

Who knows... you know, like I said, maybe there's something up with the GPU itself... some bad memory on the GPU... an eeeeeevil diode just causing problems (lol )...

...guess I'm only going to find out through testing. Just a lot of hassle to go through right now, especially given that I don't have Win 7, not that I (and some others) think that's even the problem. As stated, I really don't think AMD would have missed that in R&D, heh.

OK, the hell with it... I'll try a fresh install on my other drive, the newer ATi drivers from AMD's site, install some games and give it a whirl.

I might have already installed nVidia's drivers on that drive though... damn... grr.

Ghettobox said:

Yeah for benchmarking it is. Then again I am going to guess he's not running some gutted version of XP, probably has some services/prog's running in the background etc. I think we are looking at real world performance, on a real world machine. With 2gb of ram, and 1gb on the card he is at 3gb of a theoretical 3.5gb limit for a 32bit OS. I have no idea what kind of optimizing he has done, or hasn't done but the machine is pretty much at the limit for xp32 bit. I still would contend most gamer's have moved to Vista/7 and it's going to make his problem all the harder to track down. And generally minimum spec's are just that when it come's to games/hardware. We all know the more you can exceed them the better the real world experience will be. I am not saying it's the issue (xp, or 2gb of ram) but there is a mountain of evidence that point's that way, I'm really not sure how any other stance can be defended.

Click to expand...

Heh, actually, no... you'd be wrong on that. It is very much "gutted". I "gut" everything, and have almost nothing running the background.

As I've said, I'm not a novice regarding system building, and I gut the hell outta XP. I have about... 16 or so max processes running total in the bg ...that's including nVidia's CP etc.

...and, like I said, maybe it's the GPU itself... never know.

But, I'll give a shot. My other drive has a clean and "gutted" XP 32 installed on it.

I just cant imagine it being the OS... really cant... makin' my damn head hurt.

...Why haven't you tried a fresh Windows(XP, 7, or otherwise) install yet?

Your XP install may just be fscked, don't want to try Windows 7, fine, don't, but PLEASE, for the love of god, TRY A FRESH CLEAN FORMATTED HDD/SSD WINDOWS INSTALL.

Theres not much more advice we can give you till you try that.

Slachtbeest said:

Plus, I really don't think that, regardless of how powerful these GPU's are, that OC'ing the CPU is what's necessary to get the 5870 to run the way it should. There has got to be another issue, because the E8600, while no longer the newest CPU, is definitely (even at stock) the fastest non-extreme dual-core on the market and still holds it's own even at stock speeds.

Click to expand...

When you say "run how it should" I have to wonder what you're expecting. I've yet to see a review where this card was paired with a CPU "lesser" than i7 920 or Phenom 2 955. The E8600 is clearly in a different class from these CPUs (and the accompanying configurations, alway boasting 4GB or higher memory count) used to review the card - the fact that your numbers don't match up to what you're expecting might just come down to this fact - which is probably why people are suggesting that you OC your CPU to see if you get numbers closer to what they "should" be.

Run DriverSweeper, and clean up all unused drivers, before you install the OS.