-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
New Fermi performance figures, legit or troll ?
- Thread starter Kaizer
- Start date
Soldato
- Joined
- 24 Aug 2009
- Posts
- 2,931
LOL, love it dave. 
It's not launching at CES, almost impossible, shown, sure, launched with specific specs we can believe seen on slides that are official, almost 0% chance of that. Might we here the 380GTX is a 512sp part at 2Ghz, sure, might that become a 380gtx ultra mega can't buy me EVGA part by release, sure.
Its trolling without question but the load temps aren't anything out of the ordinary, its a power monster sure, but temps are based on concentration of that power on the size of the core. the 58XX series uses less power, sure, and probably a lot less leakage, but its also 2/3rds the size, maybe even a little less than 2/3rds.
You have more heat to dissipate, but a 50-60% larger surface area its being generated over and can pass that heat onto the heatsink. Considering my 5850 overclocked runs pretty damn cool, I have no real reason to believe a GF100 will run at 462C load, its completely possible for it to use twice the power and run cooler. Power and temperature aren't directly correlated, there are a tonne of variables and all the things that would help reduce the temp, like massive, massive, massive Arizona sized core, work in Nvidia's favour there.
But its pretty simple, its NOT faster than a 5970. If, and its a big if, they said it was 8% faster thana 5970 and left it at that, easily, there will probably be one two Nvidia made games that will work better on Nvidia hardware just enough to pull that lead, on average, it will without question be slower, significantly. Same goes for the 5870, it won't be 48% faster on average, in a couple games, maybe, in everything, no.
Ofcourse, again average could be the average over Nvidia's best games over the past 4 years, and of course, Batman AA with physx enabled, infact those numbers are perfectly possible if Nvidia only went with physx enabled games.
The biggest issue we'll have is that Nvidia have a reputation for releases of late, where they only give cards to those that sign NDA's with clauses forcing them to ONLY run very specific benchmarks, which can significantly alter the outlook of any card.
Of course the single most obvious part, the GF100 will perform 40% faster than a 295GTX, but cost less............. lol. Sure it will be sub £300.
Its trolling without question ? You just stated its NOT faster than a 5970, oh the irony of it all.
Could you point me to a sub-£300 gtx295 please. Also, he did mean that it will cost less than a gtx295 at todays prices didnt he, he didnt mean cost less than a gtx295 on its release ? lol.
Ofcourse, again average could be the average over Nvidia's best games over the past 4 years, and of course, Batman AA with physx enabled
Batman AA with physx enabled performs poorly
Depends on what you're running it on Duran..
I can play it at 1680x1050 completely maxed out with the highest PhysX setting, and it runs perfectly, nothing poor about the performance at all, yes there is a performance hit, but the frame rates are more than acceptable...
I can play it at 1680x1050 completely maxed out with the highest PhysX setting, and it runs perfectly, nothing poor about the performance at all, yes there is a performance hit, but the frame rates are more than acceptable...
not matter what when you enable physx ati cards easily outperform nvidia cards.
My only experience with PhysX was Mirror's Edge on my old 280....was quite unimpressed to say the least. The effects were cool but running around shooting flags and breaking glass gets old very quickly. Also the weird physx fog stuff realllyy slowed things down.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Batman AA with physx enabled performs poorly
Performs poorly, but far better on Nvidia hardware than AMD, not quite sure how that wasn't obvious and easy to understand.
As for Gills, trolling, irony, so you're saying an architechture that is most certainly slower transistor for transistor, will somehow suddenly beat a higher power AMD part with two cores that it has not done so for 2 generations?
A 3billion transistor 380gtx will not beat a 4.1billion transistor 5970, it simply won't happen, there are 3 fanboys on earth who believe it will happen, the majority of Nvidia fanboys think its a ludicrous claim and anyone whose not a fanboy never even once considered the possibility. It can not beat it, the 285gtx came no where near a 4870x2, we're looking at, with both the 380gtx and the 4870x2, a roughly doubling of transistors with very minor architechture changes.
first result on google, the page before shows a 4870 beating a gts 250, that page shows it being utterly demolished by enabling some crap effects. http://www.tomshardware.com/reviews/batman-arkham-asylum,2465-6.html
http://www.hardocp.com/article/2009/10/19/batman_arkham_asylum_physx_gameplay_review/9
here the last two graphs at the same settings show a single 5870 getting a min of 1, and average of 13, and a gtx 275 utterly destroying it from the graph below.
Last edited:
ati cards cant run physx. its like comparing dx11 unigine benchmark result with a dx10 one and trying to interpreter the results. physx without an nvidia card runs on intel hardware
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
I've just realise and assume he thinks Nvdia + physx = worse than AMD without physx.
AMD beat Nvidia flat out on performance in this game, without question, enable Physx on BOTH cards and AMD take a hugely bigger hit, because its doing physx on the cpu, however Nvidia have gone around for months saying its because there cards are simply faster. Remember around launch when Nvidia were saying essentially "meh, our 9600gt is faster than a 5870 in batman AA with physx enabled" .
Also a 5870 with a dedicated nvidia gts250, compared to a Nvidia card with a dedicated physx card gets soundly spanked aswell, the 5870 gets 1fps less than SIX TIMES higher min framerates, which will make a massive difference in game.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
what, physx without an Nvidia card can only be run on Intel cpu's? whats that got to do with.... anything whatsoever.
If you enable physx in game, as BOTH links showed theres a massive hit on the 5870/4870 framerate, and a big hit also on Nvidia, however it also causes Nvidia cards to be significantly faster, I'm not sure what your problem is, this is VERY old news, that every single person(bar you) seems to know.
Nvidia have already shown benchmarks of the 5870 getting beaten with physx enabled because AMD hardware sucks according to them. I obviously am not trying to say Nvidia are faster, or that a 5870 sucks, I'm pointing out most benchmarks can be skewed to give any data you want.
In this case the performance figures in the OP could be true, it just depends what Nvidia run, how they twist the results and if anyone lets them get away with it.
Both companies have done this before, infact everyone has, you run the game that runs best on your hardware(or software that runs best on your soundcard, cpu, gpgpu, hard drive, etc, etc) and make the results look rather different to real life performance.
Caporegime
- Joined
- 8 Jul 2003
- Posts
- 30,063
- Location
- In a house
Temperature might be right, got to be some reason why those cases they need are £150+
Some of you guys have probably seen this already, I just stumbled across it when I was looking for information on fermi..
I found it pretty funny!!
LMAO
mathwat
If all they have benchmarked is DIRT2, crysis and AvP3 then these are probably about correct... the 380 specs put it around 40-45% faster than the 360 which is intended to go head to head with the 5870... and crossfire scaling isn't great in most of these titles - averaging about 34% scaling at 1920x in dirt2 and AvP3 - not sure crysis off hand... dunno if they mean a cryengine 2 tech demo (in which case it probably don't even have a proper crossfire setup) or warhead.
Last edited:
