I'd be disappointed if they didn't and if they did then better overall results have nothing to do with drivers but with game selection.
AMD Vega Hardware Reviews
- Thread starter ArkeoTP
- Start date
Initial review:
Today:
http://techreport.com/news/32779/take-a-sneak-peek-at-our-geforce-gtx-1070-ti-results
RX 56 is now faster than 64 initially was. The speedup is close to 20%. So there, magic drivers really are a thing after all.
Not even remotely magic enough to threaten the 1080 Ti, however. Well, not yet, anyway.
Eyeballing this graph, it looks like V64 went from mid-high 50s to mid-high 60s while 1080 went from low 60s to high 60s. Magic drivers for all!
Ah, so you just posted something misleading. I see
Let's just wait until we see what games are used and judge afterwards.
D
Deleted member 2197
Guest
Judge on fps per dollar?
DavidGraham
Veteran
It's a fps per dollar chart, any changes in price will skew results. Besides, tests drom PCGH, ComputerBase, TPU, and others indicate nothing of this sort.RX 56 is now faster than 64 initially was. The speedup is close to 20%. So there, magic drivers really are a thing after all.
The FPS per dollar chart has that "FPS" column there, you know? You can ignore the "per dollar" portion if you want to compare just performance.
TPU doesn't use the latest drivers on AMD, but if we use R9 390 as "anchor", since it's using same really old set of drivers on both reviews, we can see that at 1440p Vega 64's performance was 67% better than R9 390 on launch, now (17.10.1, not 10.3) it's 72% better, so even they report clear improvement (not enough for Vega 56 to catch Launch64 but still).
edit: on 1080p Vega 64 has gone from 59% faster to 66% faster, on 4K from 69% faster to 75% faster
DavidGraham
Veteran
So yes the expected 5% improvements, nothing to sneeze at, it's probably in titles where Vega 64 was a 1070 like performance, don't see 20% there though, that was the main point.
You also can check ComputerBase or PCGH for up to date drivers.
You also can check ComputerBase or PCGH for up to date drivers.
Unless the original is modified with the 3 games dropped then there's nothing to talk about, the comparison is useless.
Well my point is any comparison using these charts over time is useless even with the OP as no-one knows what changed this time until the full review is out; I was just showing how they dropped 3 games in the space of 3-4 months and that increased the fps on the same charts used by the OP by around 10fps for the 1080ti.
So it is too early to say it is driver improvement for Vega as some feel.
Otherwise if we did not know 3 games were dropped between the 1080ti review and the Vega review we would assume the 1080ti managed that 9-10 fps improvement just from drivers and game patches.
Not if you only look at the performance axis.
DavidGraham
Veteran
So? The graph still shows GTX 1080 faster than Vega 64.
At any rate, it's pointless to argue about it, because they indeed changed the games selection, they added Forza 7 and Wolfenstein 2.
http://techreport.com/review/32766/nvidia-geforce-gtx-1070-ti-graphics-card-reviewed/3
If we want a more fixed point comparison, I suggest looking at Hardware.fr reviews, they use a test lineup that heavily favors AMD, Vega doesn't advance much on the 1080, and in fact goes slightly backward.
Launch Vega: http://www.hardware.fr/articles/968-17/recapitulatif-performances.html
Present Vega: http://www.hardware.fr/articles/971-17/recapitulatif-performances.html
Last edited:
We did an interesting test at io-tech, one I haven't seen elsewhere. We pitted up OC'd GTX 1070 Ti vs undervolted Vega 56
https://www.io-tech.fi/uutinen/paiv...lotettu-gtx-1070-ti-vs-alivoltitettu-vega-56/
In Battlefield 1, where stock GTX 1070 Ti barely beat stock Vega 56, undervolted Vega 56 actually ended up faster than OC'd GTX 1070 Ti.
In RoTR the story wasn't as pretty though, as undervolted Vega 56 only passed stock GTX 1070 Ti
https://www.io-tech.fi/uutinen/paiv...lotettu-gtx-1070-ti-vs-alivoltitettu-vega-56/
In Battlefield 1, where stock GTX 1070 Ti barely beat stock Vega 56, undervolted Vega 56 actually ended up faster than OC'd GTX 1070 Ti.
In RoTR the story wasn't as pretty though, as undervolted Vega 56 only passed stock GTX 1070 Ti
So when custom Vega cards come out should mean trading blows with custom 1070ti when taking in broad range of game.
I do think Nvidia messed up with their pricing and has not taken that into account, I suppose Nvidia can drop the MSRP but looks kinda silly doing that so soon if AMD manages to sell those custom models hopefully earlier rather than later.
Anyway I do not think the current price of 1070ti is that enticing, fingers crossed AMD partners do not end up just matching said Nvidia price with custom models using same cooling and similar performance between them, some have done that in the past
MSI and Asus come to mind, but at least there is XFX and Sapphire to provide some price sanity with some models.
I do think Nvidia messed up with their pricing and has not taken that into account, I suppose Nvidia can drop the MSRP but looks kinda silly doing that so soon if AMD manages to sell those custom models hopefully earlier rather than later.
Anyway I do not think the current price of 1070ti is that enticing, fingers crossed AMD partners do not end up just matching said Nvidia price with custom models using same cooling and similar performance between them, some have done that in the past
MSI and Asus come to mind, but at least there is XFX and Sapphire to provide some price sanity with some models.
Bondrewd
Veteran
Custom models won't do anything, heck, STRIX 64 managed to perform worse than the reference card.So when custom Vega cards come out should mean trading blows with custom 1070ti when taking in broad range of game.
I do think Nvidia messed up with their pricing and has not taken that into account, I suppose Nvidia can drop the MSRP but looks kinda silly doing that so soon if AMD manages to sell those custom models hopefully earlier rather than later.
Anyway I do not think the current price of 1070ti is that enticing, fingers crossed AMD partners do not end up just matching said Nvidia price with custom models using same cooling and similar performance between them, some have done that in the past
MSI and Asus come to mind, but at least there is XFX and Sapphire to provide some price sanity with some models.
Something is up with the Vega64 in terms of ease to OC, that site manually OCing managed to get the Vega 56 comparable to the manual OCd Vega 64, albeit just testing two games; my perspective is it is comparable from a marginal perspective I agree but shows it was substantially easier to OC the Vega56.
https://www.io-tech.fi/artikkelit/testissa-amd-radeon-rx-vega-56/
https://www.io-tech.fi/artikkelit/testissa-asus-rog-strix-rx-vega-64-oc-edition/
Also worth noting the Vega 56 OC seems to be marginally higher than the undevolt vs 1070ti, but probably makes more sense to reduce heat rather than go for the extra 1-3% performance - this figure is skewed a bit it seems either by patches or drivers if you look at the base performance in the 1070ti and the Vega56 review as the original review has lower base performance for Battlefield1 but higher base performance for RoTR.
One would assume the base parameters would not change between reviews and tests so driver/patches influenced *shrug*
Edit:
Unfortunately io-tech do not help that they do one review of STRIX Vega 64 at 1440p and the reference Vega 64 card at 4k.
Last edited:
Similar threads
