8600 does have 32 Stream processors

Unless he has a SFF PC like I do, then dual slot cards mean sacrificing a PCI slot or modding the side of the case.

On top of that alot of people dont like messing around with MIR's personally i do not take MIR into account when im look for something. Hell if i did i would shop at compUSA their whole damned ad is MIR's wich drives me nuts.
 
Hehe, I've stopped taking MIR into account as well when looking at prices. I wish they'd stop doing those and turn them into instant discounts. :p
 
Still waiting to eat my words ... ;/
Me too. Most of the forums on the net are laughing at the Hard review. Every other review has reservations about this car especially when most people are trying to play at 1280 which is the native res of 17 and 19inch lcds. The Hard review is the only that seems to think the 8600gts is great.
 
All I can tell you is what I experienced, and that's in th evaluation. I actually played games with both video cards, and I found the 8600 GTS to allow a better gaming experience compared to the X1950 Pro, the facts are the facts.
 
All I can tell you is what I experienced, and that's in th evaluation. I actually played games with both video cards, and I found the 8600 GTS to allow a better gaming experience compared to the X1950 Pro, the facts are the facts.
And all I can tell you is that every other review, you know, the ones with reliable and repeatable, identical benchmarking (or timedemos, whatever you want to call them) show that the 8600GTS is a shit card.
 
And all I can tell you is that every other review, you know, the ones with reliable and repeatable, identical benchmarking (or timedemos, whatever you want to call them) show that the 8600GTS is a shit card.

I'm sorry, but timedemos and "whatever you want to call them" are not reliable gaming performance evaluations.

Our results are different because they show real-world gameplay. Timedemos do not.

If you find value in Timedemos, there are plenty of other websites for that, you won't find that here, we focus on real-world gameplay.
 
I'm sorry, but timedemos and "whatever you want to call them" are not reliable gaming performance evaluations.
They give the same results on every test. They accurately show the performance of a card when compared to the exact same situation of another. How is it not reliable?

define:reliable
Brent_Justice said:
If you find value in accuracy, there are plenty of other websites for that, you won't find that here, we focus on randomization!
I'll leave it at there.
 
I'm sorry, but timedemos and "whatever you want to call them" are not reliable gaming performance evaluations.

Our results are different because they show real-world gameplay. Timedemos do not.

If you find value in Timedemos, there are plenty of other websites for that, you won't find that here, we focus on real-world gameplay.

Sorry but rendering in a time demo is still rendering. They're not pre-rendered or anything of the sort. Benchmarking real world, which is quite nice, can lead to huge variances on each run.
 
They give the same results on every test. They accurately show the performance of a card when compared to the exact same situation of another. How is it not reliable?

define:reliableI'll leave it at there.

Perhaps you have missed our past editorials, I will link them so that you can educate yourself on why we do what we do:

http://enthusiast.hardocp.com/article.html?art=NjE2LCwsaGVudGh1c2lhc3Q=

Read that and check out the links in there to other editorials, that should help you understand where we are coming from.
 
Sorry but rendering in a time demo is still rendering. They're not pre-rendered or anything of the sort. Benchmarking real world, which is quite nice, can lead to huge variances on each run.

Timedemos in most cases do not playback gameplay physics or AI, two very important aspects of gaming. I'll use FarCry as an example, if you use the timedemo feature in that game it will not use the physics engine in real time. That right there in itself makes timedemos invalid for showing gameplay performance. Beyond that though just sitting there and running a timedemo removes the person from actually getting in the game and seeing how it feels. Simply looking at min/max/avg framerates do not tell you how the game actually plays.

Some games feel fine at 24 FPS, others need to be at at least 30 FPS. It also depends on the type of game, in a game like Supreme Commander 20 FPS is acceptable because of the type of game it is. But at what point do the framerates become unplayable or what settings can you increase before the game becomes unplayable? Timedemos cannot tell you this information.

There is no better way that I know of to find out how a video card performs in a game and what settings you can use (in-game/resolution/aa/af) other then launching the game and playing it!

If you know of another way, please, let me know.

Here is an important question, as a gamer, when you install S.T.A.L.K.E.R., did you buy it to sit there and watch scripted timedemos, or did you buy it to actually play the game through and enjoy it? I'm going to guess the latter, and our evaluation method replicates the latter.
 
Timedemos in most cases do not playback gameplay physics or AI, two very important aspects of gaming. I'll use FarCry as an example, if you use the timedemo feature in that game it will not use the physics engine in real time. That right there in itself makes timedemos invalid for showing gameplay performance. Beyond that though just sitting there and running a timedemo removes the person from actually getting in the game and seeing how it feels. Simply looking at min/max/avg framerates do not tell you how the game actually plays.

Some games feel fine at 24 FPS, others need to be at at least 30 FPS. It also depends on the type of game, in a game like Supreme Commander 20 FPS is acceptable because of the type of game it is. But at what point do the framerates become unplayable or what settings can you increase before the game becomes unplayable? Timedemos cannot tell you this information.

There is no better way that I know of to find out how a video card performs in a game and what settings you can use (in-game/resolution/aa/af) other then launching the game and playing it!

If you know of another way, please, let me know.

Here is an important question, as a gamer, when you install S.T.A.L.K.E.R., did you buy it to sit there and watch scripted timedemos, or did you buy it to actually play the game through and enjoy it? I'm going to guess the latter, and our evaluation method replicates the latter.


As I said I understand your point of view, but then I can't see how you can compare video cards to each other considering all the variables of real world gameplay. And, the subject of what fps is playable and what isn't is HIGHLY subjective. For example, I play ut competitively and I can tell drops below 70 FPS pretty easily. In many fps games the minimum fps I require is 70. Many people can get by with alot less fps and not notice a thing, so that adds even more variables into your testing environment. It is impossible for you to make sure the cards are being benchmarked under equal environments and because of the type of testing you do, it's all subjective.
 
As I said I understand your point of view, but then I can't see how you can compare video cards to each other considering all the variables of real world gameplay. And, the subject of what fps is playable and what isn't is HIGHLY subjective. For example, I play ut competitively and I can tell drops below 70 FPS pretty easily. In many fps games the minimum fps I require is 70. Many people can get by with alot less fps and not notice a thing, so that adds even more variables into your testing environment. It is impossible for you to make sure the cards are being benchmarked under equal environments and because of the type of testing you do, it's all subjective.

Yes, we do subjective real-world gameplay evaluations.

We can compare video cards because we are actually being objective about what we feel is playable in a game. We aim for equal performance and see which one allows higher settings.

No one is telling you what settings to run the cards at, we are simply telling you our experiences using them and what we found was the best playable settings. You can completely ignore the framerates and still see which card is faster. All you have to do is look at what settings we found playable with both video cards.

If Card A was playable at 1024x768 NoAA/16X AF but Card B was playable (at the same framerates as Card A) at 1600x1200 16X AA/16X AF well obviously Card B is a LOT faster.

Now, if Card A was playable at 1024x768 NoAA/16X AF but Card B was playable (at the same framerates as Card A) at 1024x768 2X AA/16X AF Card B is only slightly faster, but it would still provide the best gameplay experience because it is able to run with a higher setting, 2X AA, with equal framerates to Card A, it would be the more powerful card, but not as much as in the first scenario.

You don't even need to see the framerates to figure out which card is more powerful and by how much. But we do include the min/max/avg and a graph with framerates over time to backup our claims. We firmly believe in showing you the data that leads us to these conclusions, we have to back up our claims, and that is what we strive to present to you.

If there is any way to improve upon this processes please by all means let us know, we want to better present this information to you.

If you are looking for a review using the scientific method this is not it. We instead use a more relevant to gaming method that finds out which card is going to provide the best gaming experience for the gamer, the scientific method of old does not work with today's video cards to tell you this information.

Nothing else that I know of is going to tell you how a video card looks and feels and performs in a game other then launching the game and playing it with said video card. If you know of another way, let me know.
 
I was under the impression that physics and AI are exclusively handled by the CPU, and therefore irrelevant to a video card's performance. That is not the case?
 
It is becoming less of a case when pieces of physics are being off loaded to the GPU.

Regardless, physics and AI are part of the gaming experience, to not include them completely ignores important gameplay aspects.
 
It seems that the card is a better performer than most of us initially gave it credit for, although I do wish it had also been compared against the similarly priced X1950 XT rather than just the Pro. It doesn't seem quite right to compare factory OC'd 8600's to a stock-clocked X1950 Pro and then act like they're priced equally when the latter can be found for $145, valuing a mail-in at half its face value. That's $55 cheaper than the cheapest 8600 GTS, and $85 cheaper than the cards used in the review. I would have liked to have seen the GTS pit against the X1950 XT and the GT go up against the Pro, since that puts the cards against their closest price competitors.
 
It is becoming less of a case when pieces of physics are being off loaded to the GPU.

Regardless, physics and AI are part of the gaming experience, to not include them completely ignores important gameplay aspects.
Does STALKER offload physics to the GPU?

Also, I don't understand how a card that's faster in GPU specific benchmarks would be slower in real world game play assuming all other things are equal, and the GPU is not handling physics or AI. Is there something I can read that would explain that?
 
I think a lot of people are missing out on the fact that the 8600GTS is supposed to be "mid-range", and well, right now, it pretty much falls there.

It is a little disappointing that my x1900xt is faster, so hopefully a 8600Ultra will be released, or some kind of better ATI "mid-range" card will be released, or the 8800GTS 640MB is what I'll be looking at to hopefully become "DX10 Crysis ready".
 
Does STALKER offload physics to the GPU?

Also, I don't understand how a card that's faster in GPU specific benchmarks would be slower in real world game play assuming all other things are equal, and the GPU is not handling physics or AI. Is there something I can read that would explain that?

Benchmarks and games are two very different things that react differently. Getting 50 FPS AVG in a timedemo may not be what you actually get running through the game yourself in that same scenario.

The only way to see how a video card performs in a game is to run that game and play through it. That's what we do, and that's exactly how gamers play the game.
 
with my 2 cents.

Brent, keep up the good work.

Before the cards were released or before the reviews were even out, I said that the 8600GTS would beat the X1950PRO. It's in this thread, check it out. I knew just by the numbers and clock speeds that the thing would be faster...the technology is better and runs on shader 4.0.

I keep hearing the discussion about PRICE but people are also comparing apples to oranges IMHO. People keep taking the XFX top card in the 8600 category at 229.00 and comparing it to a BASE and rebated 8800GTS 320. Okay, I agree that a base 8800 will kick the crap out of the top 8600 card...but you're not comparing the correct versions within their category to determine value...the XFX 8800GTS 320 (factory OC) is 309.00.

Compare base cost to base cost. A 199.99 base for the 8600 GTS to a 299.00 base for a 8800GTS. That's a $100.00 difference. I hate buying the BASE model of something. Don't compare a card that is getting rebated heavily to a new release. That is like comparing last years Allen Iverson shoe to this years Lebron James shoe. Both keep your feet warm, don't they?

In a couple months there will be discounts on the 8600 series and at a lower price point it will still beat the 1950PRO. The testing didn't lie. How do you get around that??? It is still a good card.

Brass
 
I think a lot of people are missing out on the fact that the 8600GTS is supposed to be "mid-range", and well, right now, it pretty much falls there.

It is a little disappointing that my x1900xt is faster, so hopefully a 8600Ultra will be released, or some kind of better ATI "mid-range" card will be released, or the 8800GTS 640MB is what I'll be looking at to hopefully become "DX10 Crysis ready".

the 7600gt and 6600gt were mid range as well and could surpass and or outperform in some situations the previous generation's high end card. It seems that the 8600 series is not capable of doing that so people are pretty irritated with the 8600's mediocre performance.
 
In a couple months there will be discounts on the 8600 series and at a lower price point it will still beat the 1950PRO. The testing didn't lie. How do you get around that??? It is still a good card.

Brass

relative to other sites' results the testing did lie. I'm still at a loss as to how rendering pixels in one engine (time demos) isn't rendering pixels in the same engine (real world) and can spit out such different results.
 
Benchmarks and games are two very different things that react differently. Getting 50 FPS AVG in a timedemo may not be what you actually get running through the game yourself in that same scenario.

The only way to see how a video card performs in a game is to run that game and play through it. That's what we do, and that's exactly how gamers play the game.

Hell i get far less FPS in most games then i do in its time demo i stopped counting on timedemos and any type of built in benchmarks. I personally am sick of the same old canned benchs everyones spewing out

And as to pricing and such blah blah how great is the slightly cheaper X19XX when DX10 games are here? i would rather pay the premium to know that in 6 months when X game i want is out i can play it as it was meant to be played. Kinda hard to argue with the features and performance VS. Price so fucking what the X19XX is cheaper its old tech it sure as shit better be cheaper... been the same in the tech community for years... you want the new tech you pay a little more plain and simple.

People are comparing launch prices and such like they have never seen a launch before... in a month or two these cards will be cheaper further invalidating your arguments. If you really dont like the pricing write a letter to ATI and tell them to hurry the hell up and compete so we can all have cheaper shit.
 
Specs/Schemcs, Price/Performance, sure. I don't care if one card uses shaders and the other stream processors, SM3.0 or 4.0, 64MB 64bit DDR or 640MB 256bit GDDR3. I have a set performance metric with my LCD and my peepers own pixel resolving resolution. I want the most eye candy I can get with good performance at a set resolution. To only compare "midrange" cards seems to miss the point of an evaluation to me. I always want to compare the extremes, so I can gauge whether or not the midrange is worth it.

Sure we get all excited about the specs, when we think we know what they will represent. Sometimes we're right, sometimes not. I realize you cannot compare specs with dissimilar architectures. Specs only matter when comparing similar architectures. So specs are meaningless when comparing the 8600GTS with the X1950Pro, but we should consider them when comparing it with the 8800GTS.

Compare these:

11763129813hYM7R3Fqx_6_3.gif
1172339741FPItSm9He2_3_3.gif

11758703864l6otAnKB2_8_3.gif


So to me, the 8600 isn't worth the upgrade, it isn't doing 1600x1200, which is my target resolution. So right now, I would need a $300 card. The number of stream processors don't matter in that, it is just a number. But it is obvious to me from your own reviews that 32 ain't enough. I think we are only seeing 32 since there is no ATI X2600 yet.

It is hard to see apple to apples here, but it looks like resolution alone would account for a 30% improvement in performance of the 8800GTS over the 8600GTS. That isn't counting the 2xAA and all that grass which KILLS my 7600GT. So I'd hazard a guess that performance is about double in this game. Worth an extra 50% cost to ya?

Hell, the GTX really starts to pick up here on the price/performance comparison, but 19x12 is more than I need, and I don't see much advantage in AA above 4x. The GTX is more than I need, and more than I can afford as well.

I am sorry, but this looks to be a little too much enthusiasm about a so-so card to me. I guess I am just atypical. I am also only considering one game here, again, this is what matters to me.
 
I agree with the others as to disappointment in this card. I have an 8800 so none of this matters to me, but it looks to me very much like the disparity between high-end and midrange has grown (ie. taking the ratio of 8800 to 8600 vs. the ratio of 7800 to 7600), which all in all, is a bad sign. The rich get richer and the poor...well, they stay poor.

Also, Brent specifically mentioned that he does not test using the scientific method. This, I disagree with, but... if he wishes to do it that way, then so be it, so all of those that complain about that should drop the argument. I think there is no better way than the scientific method, but my opinion will nevertheless be overruled by those that don't believe in it.
 
I agree with the others as to disappointment in this card. I have an 8800 so none of this matters to me, but it looks to me very much like the disparity between high-end and midrange has grown (ie. taking the ratio of 8800 to 8600 vs. the ratio of 7800 to 7600), which all in all, is a bad sign. The rich get richer and the poor...well, they stay poor.

Uh, you do realize there are two cards between the "rich" and the "poor" right?

P-R-I-C-E-P-O-I-N-T-S
 
Uh, you do realize there are two cards between the "rich" and the "poor" right?

P-R-I-C-E-P-O-I-N-T-S
You seemed to miss the POINT. The 8600gt and gts have way way less performance than the 8800gts 320mb. There is NOTHING in between.
 
So to me, the 8600 isn't worth the upgrade, it isn't doing 1600x1200, which is my target resolution. So right now, I would need a $300 card. The number of stream processors don't matter in that, it is just a number. But it is obvious to me from your own reviews that 32 ain't enough. I think we are only seeing 32 since there is no ATI X2600 yet.

It is hard to see apple to apples here, but it looks like resolution alone would account for a 30% improvement in performance of the 8800GTS over the 8600GTS. That isn't counting the 2xAA and all that grass which KILLS my 7600GT. So I'd hazard a guess that performance is about double in this game. Worth an extra 50% cost to ya?

Hell, the GTX really starts to pick up here on the price/performance comparison, but 19x12 is more than I need, and I don't see much advantage in AA above 4x. The GTX is more than I need, and more than I can afford as well.

I am sorry, but this looks to be a little too much enthusiasm about a so-so card to me. I guess I am just atypical. I am also only considering one game here, again, this is what matters to me.

What great expectations you have of a mid-range card. So you wanted a 8600 GTS to run @ 1600x1200 with the in-game settings, set to max ? That's high-end ground, where you definitely should be looking, if you're in the market for a card and that means 8800 GTX/8800 GTS. Mid-range is 1280x1024 and below and that's where the 8600 GTS does better.
 
You seemed to miss the POINT. The 8600gt and gts have way way less performance than the 8800gts 320mb. There is NOTHING in between.

At this point, that's true, But given that NVIDIA has the GS moniker, there should be a 8800 GS/8900 GS soon enough, which will take that position.
 
All I can tell you is what I experienced, and that's in th evaluation. I actually played games with both video cards, and I found the 8600 GTS to allow a better gaming experience compared to the X1950 Pro, the facts are the facts.

serious case of tunnel vision


The 8600GTS delivers 40% of the performance of the 320MB 8800GTS for 70% of the price.

The 8600GT outperforms a 7600GT - but is priced like a 7900GT.
 
Myself, I read this review, then promptly ordered a 8800gts. (EVGA, I'll step up if I feel the need)

The 8600 doesn't have much, if any, of a performance advantage over the X1800XT I purchased last August for $200.
 
Anyone do a quick price now vs price at launch ratio for the 8800s? Im sure, in 3 months or so, the 8600gts will be heck of a lot cheaper, and thus all will be better :)
 
8600 GTS $199-$229
8800 GT 320 $259-$309

I mean honestly... Just go for the 8800. The 8600 GTS should have been named the 8500 GTS and the 8500 GT should have been the 8200 GT.

Then when the refresh comes, there won't be confusing/bunched up numbers and stuff.

Oh, and yeah, I'm pretty sure the 8600 Ultra will have 48 Stream processors @ 750 MHZ after they see what the AMD/ATI cards will do.
 
i think people were hoping that the 8 series would eclipse there prodecessors across the board regardless of class at each price level just like a single 7800gtx against 6800ultra SLI and the 8800GTX vs 7900gtx sli. I wonder how many of these people moved up in monitor resolution.

i wonder if 8600gt SLI can best my single 7950gt. If they can, ill get those when they drop to $120 each, since i got my 7950gt new for 240.
 
i think people were hoping that the 8 series would eclipse there prodecessors across the board regardless of class at each price level just like a single 7800gtx against 6800ultra SLI and the 8800GTX vs 7900gtx sli. I wonder how many of these people moved up in monitor resolution.

i wonder if 8600gt SLI can best my single 7950gt. If they can, ill get those when they drop to $120 each, since i got my 7950gt new for 240.


an 8800gts can best your 7950 at $240 and is faster than 2 8600gt's in sli, WOOO
 
yea, this card performed better then i thought, but it still falls in expectations of mid ranges, i'll have to pass on this card (reminds me of the X1600's last gen heh)

the cool thing about this if the e-tailers get the price down then they would be fairly good, but not for what htey are right now.

good review guys, it did prove you can't base everything on numbers, but their performance still isn't what I believe a mid range should be.
 
Back
Top