• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Why does GF100 suck?

Status
Not open for further replies.

Jacez44

Limp Gawd
Joined
May 9, 2007
Messages
417
When I saw the link to the review of the new GTX 465 I immediately thought "this is the chance to see the performance difference between GF100 and GT200" and I KNEW that it would suck, but this.. this is just sick.

Technical specs:
GTX 465 = 352 SP's. GTX 275 = 240 SP's.

Actual benchmarks:
GTX 465 = GTX 275

http://www.xbitlabs.com/articles/video/display/geforce-gtx-465_3.html#sect1

It's like the GTX 260 being identical in performance to the 9800 GTX.

So when you think you're getting a GTX 275 with 480 SP's.. think again!

Now I'm having second thoughts about my Triple SLi build...
 
When I saw the link to the review of the new GTX 465 I immediately thought "this is the chance to see the performance difference between GF100 and GT200" and I KNEW that it would suck, but this.. this is just sick.

Technical specs:
GTX 465 = 352 SP's. GTX 275 = 240 SP's.

Actual benchmarks:
GTX 465 = GTX 275

http://www.xbitlabs.com/articles/video/display/geforce-gtx-465_3.html#sect1

It's like the GTX 260 being identical in performance to the 9800 GTX.

So when you think you're getting a GTX 275 with 480 SP's.. think again!

Now I'm having second thoughts about my Triple SLi build...
Clock speeds.
 
The GTX 465 is most unimpressive from that review, but I missed your "Why does RV870 suck?" thread on the HD 5830. I would have remembered it. :p All the GTX 465 and HD 5830 show is that a decent GPU can be pared down to something much worse, not that you can claim the full version is terrible based on that.
 
Also fewer TMUs and less memory bandwidth.

What are you talking about?

The clock speed is the same between the GTX 465 and GTX 275.

The only difference is that the former has 46% more SP's but the same exact performance.
 
The GTX 465 is most unimpressive from that review, but I missed your "Why does RV870 suck?" thread on the HD 5830. I would have remembered it. :p All the GTX 465 and HD 5830 show is that a decent GPU can be pared down to something much worse, not that you can claim the full version is terrible based on that.

If the HD 5830 performed as well as the HD 4850, then your argument would be valid.
 
They are the same GPU. And I've proven how inefficient it is.

No, you didn't prove anything. You have re-iterated one of the problems that SOME people have with the concept of chip speed binning.... in not realizing that every generation that utilizes that concept will always have one card/chip that from the outset appears to perform badly.

Quite often these are great overclockers and/or are unlockable in some fashion.
 
And I've proven how inefficient it is.

All you've proven is that you don't understand GPU architectures. Looking at clock speeds alone tell you nothing. As other people have pointed out the 275 has considerably more bandwidth and texture throughput than the 465 which could offset any shading advantage for the latter.
 
If the HD 5830 performed as well as the HD 4850, then your argument would be valid.
Why?

The HD 5830 is even slower than the GTX 275. Your claim in this thread is that the "GF100 suck", based on the GTX 465 benchmarks running on par with the GTX 275. If the "GF100 suck" based on that performance, what does that say about the HD 5830?

Does a game run well on a new GPU because of how it performed on the last generation? Srsly twisted logic there.
----

edit: Anyways, assuming the benchmarks and specs are legit:

Code:
           GTX 465 / GTX 275
SPs          352       240
core         600       633    MHz
shader      1215      1404    MHz
memory       802      1134    MHz
bandwidth    102       127    GB/s
pixel      19500     17724    peak Mpix/s
[u]texture    26100     50640    peak Mtex/s[/u]
ROP           19        17    Gpix/s

Aha! Texture choked. The GTX 480 is the same. It has a maximum texture rate of 42000 Mtex/s. Weird design decision.
 
Last edited:
No, you didn't prove anything. You have re-iterated one of the problems that SOME people have with the concept of chip speed binning.... in not realizing that every generation that utilizes that concept will always have one card/chip that from the outset appears to perform badly.

Quite often these are great overclockers and/or are unlockable in some fashion.

I think GF has demonstrated that that's not going to be the case here.

I remember hearing somewhere that there was a linear shrink of GF100, like GF100b? That chip might overclock well... if its actually a shrink, which it cant be because TSMC dumped the 32nm half node and 28nm is too far off...

Yeah this whole fermi architecture, the 40nm process from TSMC is just to weak. When The absolute top binned release part from Nvidia has part of its die disabled (remember the GF100 is fabbed with 512 SP's) you know the company is in trouble.

I was really counting on some pressure here from Nvidia to get AMDs pricing down to where it should be, but it looks like AMD's going to keep skimming the market all the way through X-mas. Q4 2010 autta be good for AMD.
 
Too bad 5870 still loses to 4870x2.

DX11 gen1 has been a disappointment overall.
 
Too bad 5870 still loses to 4870x2.

DX11 gen1 has been a disappointment overall.

where are you getting either of those statements from? A simple google image search shows the first one as false (and most of those are using release drivers which have since been updated several times over for some big performance bumps in some cases). And as for DX11, show me a single application short of 3DM11 that uses anything from the DX11 API.
 
2qv8939.jpg
 
OP has point so stop with the nonsense. the GF100 gpus do suck in many ways. hell the 448sp 470 can barely even beat the gtx285 in many cases and in fact the gtx285 will match it on some rare benchmarks. the fact that a 352sp card can only keep pace with a gtx275 is a joke. we are getting no better performance yet we are getting the same or higher costs and power consumption compared to last gen. the butchered TMUs and fairly low memory bandwidth are certainly the main culprits.
 
A current-gen single GPU card isn't as fast as a last-gen dual GPU card... makes sense to me.

Too bad that 4870x2 doesn't support DirectX 11 :rolleyes:
can you even do basic math? a 5870 with the SAME number of sp, tmus, and rops along with HIGHER clocks and NO crossfire to contend with losing to the 4870x2 makes sense?
 
Last edited:
OP has point so stop with the nonsense. the GF100 gpus do suck in many ways.
If there's any primary cause to it, jen-hsun huang's single mindedness to pursue GPGPU (while falling behind to boot :p) over more focused graphics processor designs has made this mess. He's a very technical CEO and the blame does fall on him. Really, nvidia has some real problems at the top that aren't going to be fixed any time soon. The arrogance from so many years of success hasn't helped.

But that's not really what this thread is about. Gimp a GF100 enough and it runs on par with a GTX 275. Oh the horror. Gimp a RV870 enough and it runs even slower than that.
 
you're under the impression that the SPs on the GTX 4xx are the same as the SPs on the GTX2xx, which is wrong,

and another point, did you expect EVERY fermi based card to beat EVERY gtx2xx card? you know there will be a point when the new "mainstream" will compete with the older upper end cards.

Both the GTX480 and 470 are > then the GTX285/275, so the fact that the GTX465 is ~ to the GTX275/285 is not surprising, or bad, it will also be priced similarly, but now you get DX11, and supposedly better cuda performance thanks to the new shader design
 
If there's any primary cause to it, jen-hsun huang's single mindedness to pursue GPGPU (while falling behind to boot :p) over more focused graphics processor designs has made this mess. He's a very technical CEO and the blame does fall on him. Really, nvidia has some real problems at the top that aren't going to be fixed any time soon. The arrogance from so many years of success hasn't helped.

But that's not really what this thread is about. Gimp a GF100 enough and it runs on par with a GTX 275. Oh the horror. Gimp a RV870 enough and it runs even slower than that.
well on paper it doesnt look that gimped compared to the gtx275 and I believe that is the OPs point. you would certainly think that 352sp would handle games a little better than 240sp even if other parts of the 465 are a little lacking. I am just aggravated that we cannot get a better performing card for the same amount of money as we paid for the previous gen. hell I cant even buy a card for 200 bucks that soundly beats my gtx260 which I paid only $185 for 18 months ago and it came with a game. not to mention that a gtx260 was even down to below $150 with rebates.
 
you're under the impression that the SPs on the GTX 4xx are the same as the SPs on the GTX2xx, which is wrong,

and another point, did you expect EVERY fermi based card to beat EVERY gtx2xx card? you know there will be a point when the new "mainstream" will compete with the older upper end cards.

Both the GTX480 and 470 are > then the GTX285/275, so the fact that the GTX465 is ~ to the GTX275/285 is not surprising, or bad, it will also be priced similarly, but now you get DX11, and supposedly better cuda performance thanks to the new shader design
so higher price point, more power consumption yet equal performance is acceptable? the gf100 cards are almost a joke when all factors such as heat, noise, cost and power consumption are considered. the gtx465 looks even sillier than that abomination known as 5830.
 
I guess both the 5830 and GTX 465 go to show you can't resurrect a die with severe manufacturing defects but you can polish it into a turd. :p
 
so higher price point, more power consumption yet equal performance is acceptable? the gf100 cards are almost a joke when all factors such as heat, noise, cost and power consumption are considered. the gtx465 looks even sillier than that abomination known as 5830.

What is your point? the 465 is a MAINSTREAM card, it's not meant to topple everything that the GTX2xx line brought out, it's mean to replace the previous mainstream card, and it does, it's priced almost exactly where the GTX275 was too, $279 and the price will drop as usual.

The GTX275 is barely slower then the GTX285, which was the fastest "single" gpu and sold for over $400 now the GTX465 should perform about the same, and sell cheaper, why are you complaining? on that note, the GTX465 is faster then the GTX275, the only game the GTX275 was "faster" or "matched" the GTX465 was in RE5 @ 2560x1600 and ME2 @ 2560x1600, everything else the 465 was faster in.
 
Just wondering, would you guys be saying this if it sold for ~$150?
 
Last edited:
What is your point? the 465 is a MAINSTREAM card, it's not meant to topple everything that the GTX2xx line brought out, it's mean to replace the previous mainstream card, and it does, it's priced almost exactly where the GTX275 was too, $279 and the price will drop as usual.

The GTX275 is barely slower then the GTX285, which was the fastest "single" gpu and sold for over $400 now the GTX465 should perform about the same, and sell cheaper, why are you complaining? on that note, the GTX465 is faster then the GTX275, the only game the GTX275 was "faster" or "matched" the GTX465 was in RE5 @ 2560x1600 and ME2 @ 2560x1600, everything else the 465 was faster in.
I just told you what the point was. I guess I will cap and bold it for you. HIGHER POWER CONSUMPTION, HIGHER COSTS while BARELY MATCHING the gtx275 from last gen. and prices likely wont be dropping anytime soon like on the previous gen. hell ATI has only increased their prices in the last 6 months. the consumer is not getting a decent deal at all compared to last gen.
 
If this temporary band-aid card made from GTX 470 reject GPUs costs more than $200, there's just no defending it. It would be impossible to recommend it over a 5850 or even a 5830.
 
Bearly matching the gtx 275 performance? hmm, some of you really need to go and read the review again, also point out the fact that it actually give gtx 275 good whooping in some games, some games may be there is architecture limitation or may be just drivers are not on par. But to speak one side of the story is just preaching hate for a card that is not even released yet, and was tested with older drivers.

Wait and see the reviews from other websites, I am not a fanboy I jump camps all the time, but to flame a card that clearly leads in some games by a fair margin is just not fair at all.

not to mention it was pretty close to hd 5850 in dirt 2, and almost doubled the fps on gtx 275 in battleforge. yes in some games it is not much faster, but to say it is not anything faster then gtx 275 is just plain wrong, depends on the game you are playing.

if it is priced decent it might be a hot buy.
 
Last edited:
Bearly matching the gtx 275 performance? hmm, some of you really need to go and read the review again, also point out the fact that it actually give gtx 275 good whooping in some games, some games may be there is architecture limitation or may be just drivers are not on par. But to speak one side of the story is just preaching hate for a card that is not even released yet, and was tested with older drivers.

Wait and see the reviews from other websites, I am not a fanboy I jump camps all the time, but to flame a card that clearly leads in some games by a fair margin is just not fair at all.
no, YOU need to look at the review again. the gtx465 is barley faster overall than the gtx275. you can also go back and look at the faster gtx470 review to see that in some cases its no better than a gtx285.
 
Luckey777 said:
Just wondering, would you guys be saying this if it sold for ~$150?
"Awesome." :p

But at $300, "total fail."

as per wafer area / die area (assumes no chips lost from edge clipping): 2pi * (150mm)^2 / (23mm)^2 = 267 chips. If you assume 70% of those are usable that gives less than 190 usable chips. That's abominable.

Absolutely no way a part like that is going to be anything less than $250 unless someone is willing to take a loss.

yeah this is very disappointing: poor engineering on Nvidia's part (they wanted a monster, and they got one), poor manufacturing on TSMCs part, and an AMD that desperately needs money (more than market share) means this back to school season graphics card market is going to suck, and unless Nvidia and TSMC can boost yields faster than a hobo jumps on a Christmas turkey then Christmas isn't going to look much better.
 
Last edited:
but... 5830 isn't a cut down 5870...

:|

You're joking, right? Because it is.

Look, all the 460/465 benchies out there right now are simply conjecture. WE don't know if this is going to be a reference-heavy or AIB-heavy card. We don't know how they overclock.

Not only that but performance on paper vs real world gaming experience can sometimes differ. For instance I'm writing a Sapphire HD 5830 review, and despite the disagreement of others (who don't own the card and haven't used it), I FOR SURE notice a host of improvements over my previous cards (overclocked Sapphire 4580 and overclocked PowerColor 4870). Let's just say that at this point, I'm pretty well certain that HD 5830 is positively worth $209 once overclocked, and it's fairly reasonable to assume that the GF100 based 465 will feel worth the money too.

And we don't know if this (as well as the 5830) are the 6800NU/GS of this generation. Remember if you will the shortcomings of that architecture really stood out at stock speeds, but AIB partners made nicely overclocked versions, and unlocking was possible in some cases.

I believe that soon enough many 5830s will be unlocked; it's just going to take some more work by RBE and other BIOS editors. Same goes for nVidia's equivalent card.
 
Wow, how could so many people fail to understand something so simple?

352 vs 240; 352 = 240; WTF

There. It's just that simple. No, don't try to make it seem complicated to make yourselves feel better. It's just simple. Just admit you don't know.

And no, the texture units don't have anything to do with it. If it did, the GTX 480 would perform as well as the GTX 275. Unless the GTX 465 has 4 TMU's, that is irrelevant.
 
IIRC Xbitlabs used an early beta driver. Might as well wait till Computex for the full review.

It'd better be priced competitively with the 5830, else it's a goner for sure.
 
See 5770 vs 4870 - the former having more shadering but less bandwidth. Not sure how it performs these days but at the time people were disappointed with its performance vs the 4870. New architectures always seem a tad slower than they ought to be. My mates gtx260 is only maybe 20-30% faster than my 8800gt with almost identical subsystems, on paper it should be more.

Just seen that my (zotac) 8800gt has 90% as much texture fill rate as a gtx 480 lol.
 
When I saw the link to the review of the new GTX 465 I immediately thought "this is the chance to see the performance difference between GF100 and GT200" and I KNEW that it would suck, but this.. this is just sick.

Technical specs:
GTX 465 = 352 SP's. GTX 275 = 240 SP's.

Actual benchmarks:
GTX 465 = GTX 275

http://www.xbitlabs.com/articles/video/display/geforce-gtx-465_3.html#sect1

It's like the GTX 260 being identical in performance to the 9800 GTX.

So when you think you're getting a GTX 275 with 480 SP's.. think again!

Now I'm having second thoughts about my Triple SLi build...

TMUs and ROPs are diffrent too. Also slower clocks and diffrent bandwidth..
 
Status
Not open for further replies.
Back
Top