NVIDIA GF100 & Friends speculation

Nope, I think you're just reading what you want to read. We haven't really learned anything more about GF100 since the whitepaper was released and it's clear that people are still fishing. So yes he gets the benefit of the doubt for being the least wrong so far. I'm still waiting for your nomination btw.
False claim!

We now have ONE benchmark showing a (somewhat relative) heavy advantage of GTX480 over HD5870 :rolleyes:

Previously, we only had a short portion of this same benchmark, that's quite different ;)

One thing is strange though, it's the only 99% credible benchmark nVidia gave us but at the same time there's another one about 90% credible showing an ~8% advantage for the GTX470 in a less biased scenario using the same benchmark...

If performance gap is the same between 470/480 and 275/285 (7-12% depending on settings, which seems in line with both the ~8% advantage in 1920 4x/16x for the 470 and the ~30% advantage in 1920 1x/1x for the 480), it's going to be worse than RV770 vs GT200 as this benchmark is quite a best case scenario with Cypress struggling with the massive tessellation involved.

In other words, 470 would only be on par with 5870 when it struggles while 480 would be very close in actual games using more/heavier shaders and textures.
 
Last edited by a moderator:
I think we've filled our quota on debating random people on the internetz for the day, focus on gt300 (there's only two weeks left!). ;)

edit: too many posts to delete, I gave up after awhile; don't be offended if your post was deleted and someone else's wasn't (because in all likelihood, that post should probably have been deleted too)
 
I think we've filled our quota on debating random people on the internetz for the day, focus on gt300 (there's only two weeks left!). ;)

edit: too many posts to delete, I gave up after awhile; don't be offended if your post was deleted and someone else's wasn't (because in all likelihood, that post should probably have been deleted too)
You just took all the fun out of silly season.. :cry:
 
False claim!

We now have ONE benchmark showing a (somewhat relative) heavy advantage of GTX480 over HD5870 :rolleyes:

Previously, we only had a short portion of this same benchmark, that's quite different ;)
Actually, what we have is an artificial benchmark, where the Geforce has quite an advantage in short part, translating into a medium advantage over the whole benchmark. But the price seems to be a much larger chip and - according to rumor mill - a lot more power consumption.
 
We don't know yet whether we'll see 5870's doing 1.1GHz on air being sold as a separate SKU, let alone a AIB OC'ed part.

But yeah, they'll be great, especially with newer drivers.

CJ mentioned that some 5870s "Lightning" could do that, so Cypress already has more overclocking potential than RV790, apparently.
 
CJ mentioned that some 5870s "Lightning" could do that, so Cypress already has more overclocking potential than RV790, apparently.

Almost every Cypress can do 1050 MHz with raised vgpu (1,35V) and 100 % fan speed with AMD's reference cooler. Best reference card I've tested could do 1080 MHz with 1,385V. I've tested couple..

X-2010012315510694716.jpg


:cool:
 
It will be interesting what hardware review sites recieve samples and what 'benchmark suite' is going to be used. I have little faith in 'honest' reviews except from certain sites this being one of them, it is so very obvious on what sites are on the take, I remember back to the FX 5800 days where many sites did not include screen shots in their reviews, or used specific scenarios to ensure the product looked great.

If the Gtx 400 series is a good product then hopefully there won't be a need for selective reading and consumers being ripped off by marketing and reviewers on the take.
 
Wow, you're expecting a B3D review? That's the most optimistic thing I've heard about Fermi yet :LOL: I would love to be pleasantly suprised but at least hardware.fr should deliver the goods that matter.
 
Almost every Cypress can do 1050 MHz with raised vgpu (1,35V) and 100 % fan speed with AMD's reference cooler. Best reference card I've tested could do 1080 MHz with 1,385V. I've tested couple..

X-2010012315510694716.jpg


:cool:

So YOU'RE to blame for the shortages! :devilish:
 
http://www.semiaccurate.com/2010/03/12/heaven-20-benchmark-shown-gdc/

It will be interesting to compare how 1.0 and 2.0 scale across both ATI and Nvidia hardware. 1.0 scales really well on Nvidia hardware, but that performance does not translate into real world gaming performance, a typical weak point with synthetic benchmarks.

How does charlie know the Fermi scales well in Unigine 1.0? The leaked results is actually pointing towards Unigine 1.1.
 
Wow, you're expecting a B3D review? That's the most optimistic thing I've heard about Fermi yet :LOL: I would love to be pleasantly suprised but at least hardware.fr should deliver the goods that matter.

Make it be available in retail and we'll review it as soon as possible(post-launch, of course):smile:
 
Back
Top