Then blame AMD for not sponsoring more AAA titles.
But yes, they benching Hitman in DX11 is fail when DX12 runs faster for AMD GPUs. I see a lot of sites run RotTR DX12 when it runs slower for everyone.. -_-
Of course. Binning is out of question if you are offering a commodity product.
Perhaps this might be interesting. As you see, a 65 delta in measured power. I guess techalter would be around the Hardocp reading.
From 53W [Hardocp] less than a GTX970 system to 12W more [guru3d]. A 65W measurement spread for a 150W card is HUGE and can't be explained away with traditional reasons.
hardwarecanucks get 24W less
techreport gets 2W more
anandtech gets 12 less
Hardocp gets 53W less
computerbase.de gets 26W less
all of the above use total system power
techpowerup gets 7W more
Toms gets 7W less
guru3d gets 12W more
these do card power
so reference amd gpus go up against custom nv AIBs in reviews, while reference and custom AIB gpus goes up against reference amd gpus?It's because a lot of sites test against custom 970s, which aren't all the same. Overclock3D for example test it against a Zotac 970 Amp Extreme, which is faster than a stock 980 and uses a lot more power.
that's the one one review that I linked because it was most recent on my mind.
There are others, but you need to go through the reviews. I ain't gonna do it and relink them here. There is an entire thread for it
There is some consistency in the power issue being an issue, but the degree of the issue is actually rather inconsistent. You're probably also dealing with different methods of measurement here, I dunno.
The takeaway is that it is both known that reviewers largely received physically different boards than those available for retail, drivers are new with a new uarchitecture and a new software suite. Too much inconsistency to confirm the degree ofthe problem, whether or not it actually is a problem, and how it can be fixed, if it can.
Even Kyle at [H], who apparently gets a chub for hating on AMD, reports sources that claim there is a board lottery going on. Some are great, some not-so-great. Regardless, it remains true that there isn't anything on the market that can beat 480 for $200 in any measurable variable, power efficiency or otherwise, and there won't be for some time. 1060 isn't going to be released less @ than $250, I imagine.
Also: it's been 2 days ffs.
It's because a lot of sites test against custom 970s, which aren't all the same. Overclock3D for example test it against a Zotac 970 Amp Extreme, which is faster than a stock 980 and uses a lot more power.
Perfect is a goal you strive for. It is not something you typically achieve. Holding off for a perfect launch is a great way to never actually launch. There will be hiccups, always. To suggest that "AMD should have went for a perfect launch" as if that's a tangible goal for any organization, as if that's something you can just do, is incredibly naive.
You'd think all the fancy power management slides on Polaris and other products would amount to get this problem solved, yet we get results like computerbase's. They'll have to eventually get this solved because the average user isn't going to be installing afterburner or using wattman to dial voltage down and stress test to find the perfect spot, reap the savings and get the product as it more or less should be from the start.
Either the rumor of power management not enabled on RX480 is true when we get RX470 and see the 2.8x efficency number shown on the slides (RX480 isn't hitting that number), or the power management features are enabled and this is the best it can do to get voltage down to the sweet spot.
if 480 AIB can equal or beat 980 performance, it will become a 240$ mid range card, no longer a mainstream part. it will become king under 300$Joker Productions
AMD RX 480 (1350mhz) vs GTX 970 (1500mhz) Review | Benchmark Showdown
RX 480 OC wins in 8 out of 13 games; and completely smokes the GTX970 OC in all DX12 games in the review. GTX970 is basically DOA now since it has less VRAM, non-existent DX12 support/performance and has only 3.5GB of VRAM. Long-term performance-focused driver support for RX 480 vs. GTX970 should also be a no contest.
Joker noted that the card could NOT maintain 1266mhz boost out of the box in games at stock Power Limit settings. This is very disappointing since it means RX 480 reference is a repeat of R9 290 reference but for a different reason = reduced GPU clocks due to the power cap (instead of temperature thermal throttling). It's pretty shocking how AMD aimed to have an HD7850/7870 Pitcairn linearge successor but the reference card draws close to 170W. They better shift Vega to TSMC.
The best advice is the same as has been for many generations: just skip the horrid reference blower GTX1070/1080/RX 480 cards in favour of AIB versions, unless you will be manually undervolting and/or putting waterblocks/AIO on them.
The AIB RX 480 should fix almost all of the key issues related to the RX 480 reference card, aside from the mediocre perf/watt. We could potentially see a 8-10% increase in performance from AIB cards knowing that a reference RX 480 actually GPU clock throttles. This is actually a big deal since that could be enough to get an AIB RX 480 close to GTX980.
Sweclockers Summary Charts
It's shocking what a pile of garbage the GTX780Ti has become (83% vs. 100% at 1440p).
This is what I'm looking for basically. Good stuff now I hope some of the third parties will pull through with decent coolers and a nice overclock. Not like I enjoy overclocking too much yet if the custom models come with some sort of boost it will be nice. Now I just hope they release soon and that I'll be able to grab one.Joker Productions
AMD RX 480 (1350mhz) vs GTX 970 (1500mhz) Review | Benchmark Showdown
RX 480 OC wins in 8 out of 13 games; and completely smokes the GTX970 OC in all DX12 games in the review. GTX970 is basically DOA now since it has less VRAM, non-existent DX12 support/performance and has only 3.5GB of VRAM. Long-term performance-focused driver support for RX 480 vs. GTX970 should also be a no contest.
Joker noted that the card could NOT maintain 1266mhz boost out of the box in games at stock Power Limit settings. This is very disappointing since it means RX 480 reference is a repeat of R9 290 reference but for a different reason = reduced GPU clocks due to the power cap (instead of temperature thermal throttling). It's pretty shocking how AMD aimed to have an HD7850/7870 Pitcairn linearge successor but the reference card draws close to 170W. They better shift Vega to TSMC.
The best advice is the same as has been for many generations: just skip the horrid reference blower GTX1070/1080/RX 480 cards in favour of AIB versions, unless you will be manually undervolting and/or putting waterblocks/AIO on them.
The AIB RX 480 should fix almost all of the key issues related to the RX 480 reference card, aside from the mediocre perf/watt. We could potentially see a 8-10% increase in performance from AIB cards knowing that a reference RX 480 actually GPU clock throttles. This is actually a big deal since that could be enough to get an AIB RX 480 close to GTX980.
Sweclockers Summary Charts
It's shocking what a pile of garbage the GTX780Ti has become (83% vs. 100% at 1440p).
I think you're reacting as many in assigning a cause without any clear investigative thinking.It's because a lot of sites test against custom 970s, which aren't all the same. Overclock3D for example test it against a Zotac 970 Amp Extreme, which is faster than a stock 980 and uses a lot more power.
We could potentially see a 8-10% increase in performance from AIB cards knowing that a reference RX 480 actually GPU clock throttles.
Um dies testen zu können, haben wie den Speichertakt der Radeon RX 480 erhöht. Von 2.000 MHz auf 2.300 MHz. Dementsprechend steigt auch die Speicherbandbreite von 256 GB/s auf 294,4 GB/s also um 15 %.
if 480 AIB can equal or beat 980 performance, it will become a 240$ mid range card, no longer a mainstream part. it will become king under 300$
damn bacon, resize those images, there is no need for 4k images
100% confirmed by Sweclockers = out of the box, the RX 480 reference cannot maintain the 1266mhz GPU boost without a Power Limit adjustment (and most likely a more aggressive fan profile).
Joker Productions
AMD RX 480 (1350mhz) vs GTX 970 (1500mhz) Review | Benchmark Showdown
RX 480 OC wins in 8 out of 13 games; and completely smokes the GTX970 OC in all DX12 games in the review. GTX970 is basically DOA now since it has less VRAM, non-existent DX12 support/performance and has only 3.5GB of VRAM. Long-term performance-focused driver support for RX 480 vs. GTX970 should also be a no contest.
Joker noted that the card could NOT maintain 1266mhz boost out of the box in games at stock Power Limit settings. This is very disappointing since it means RX 480 reference is a repeat of R9 290 reference but for a different reason = reduced GPU clocks due to the power cap (instead of temperature thermal throttling). It's pretty shocking how AMD aimed to have an HD7850/7870 Pitcairn linearge successor but the reference card draws close to 170W. They better shift Vega to TSMC.
Of course GimpWurse is fair and neutral... get in with the program, or no more cookies.Vulkan may be more GCN favorable than DX-12.
You do know that out of 16 games in the TPU reviews the last months, 10 games are NVIDIA GameWorks and only a single one is AMD Gaming Evolved (HITMAN) and that one is benchmarked in DX-11.
Still hard to fathom, all this talk by AMD about their new power features and Wattman... and this happens:
https://www.computerbase.de/2016-06...el_potenzial_fuer_undervoltage_bei_der_rx_480
Just a simple under-volt in Wattman, they dropped power usage by 33W AND performance increased by 4.7%!
Instead of a 150W card that's behind a 390X, you now have a ~120W RX 480 that has 390X performance.
This is quite telling, because it's clear AMD's new power features are just not working as intended. It is not extracting peak optimal performance out of the chip at all.
This reminds me of the 7950 and 280 boost edition, where they just jacked up vcore to 1.25 default (earlier 7950 had 1 to 1.125 vcore) causing them to use a lot more power for nothing.
Yet another fail.
-The 970 SSC was running stock clocks (1340MHz, not the 1500MHz you put up).
-The RX480 edges out the 970 SSC only when it is overclocked to 1350MHz.
-With the RX480 at stock speeds, the 970 SSC wins in all games, but that is expected given its reference vs aftermarket.
The RX480 doesn't really smoke the 970. It barely edges it. His results falls in line with most reviews out there: both cards perform similarly.
I am not sure why you call the 970 DOA when it's been available on shelves for the past 2 years and is about to go EOL soon. You are comparing the 970 to a brand new architecture that came out 2 years later and benefits from a full node shrink. And yet, it barely edges its predecessor in performance and efficiency. The only thing it has going is its price/performance, and that is because that's the only thing AMD can do when they cant compete elsewhere. Doesn't look promising if this trend continues for Vega.