AMD Radeon R9 480 perf is Close to R9 390X

Published by

Click here to post a comment for AMD Radeon R9 480 perf is Close to R9 390X on our message forum
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
So let's say the specs of the 480 system aren't as good as the 390x one and the performance delta is closer to 20%. A 480x could definitely come in at 390x speeds, which is basically what we saw with the Hitman benchmark. A 390x @ $300 / 232mm2 isn't bad at all. If they want to compete with Pascal, they could easily just xfire them or do a dual GPU based on it.
xfire πŸ™ dual-gpu πŸ™ 390x @ $300/232mm^2 isn't bad. but considering they gave up the high end to focus on this, and considering nvidia will either launch a x60ti based on gp104, or gp106 , soon after x80/x70 i just don't see it as offering impressive value for money.. the only good thing, for them, is having the same chips both on desktop and mobile
https://forums.guru3d.com/data/avatars/m/80/80129.jpg
xfire πŸ™ dual-gpu πŸ™ 390x @ $300/232mm^2 isn't bad. but considering they gave up the high end to focus on this, and considering nvidia will either launch a x60ti based on gp104, or gp106 , soon after x80/x70 i just don't see it as offering impressive value for money.. the only good thing, for them, is having the same chips both on desktop and mobile
I agree X-fire/SLI scaling is pretty poor now, but it should technically improve under DX12 and they have hinted at making improvements to it through other means as well (AMD's Raja interview). I also know that Stardock/Oxide have plans at selling their multi-gpu component to their engine as a middleware. I don't know if their solution will catch on, but I can almost foresee a GPU Open equivalent from AMD or similar solutions from other companies. I also find it weird that AMD would essentially just concede the high end GPU market for the next 6-8 months until Vega. With a dual GPU part for $600 that might end up slightly faster than 1080 when scaling right, I could see that doing some decent sales damage to Nvidia. That being said, I think Pascal might have some hidden SLI improvements in it. Part of doing NVLink should have an effect on the way they do multi-GPU communication, similar to XDMA (I realize NVLink is server only, but it obviously effects more than just the mezzanine connector). I really wouldn't be surprised if we got a GTX1090, dual GP104.
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
I agree X-fire/SLI scaling is pretty poor now, but it should technically improve under DX12 and they have hinted at making improvements to it through other means as well (AMD's Raja interview). I also know that Stardock/Oxide have plans at selling their multi-gpu component to their engine as a middleware. I don't know if their solution will catch on, but I can almost foresee a GPU Open equivalent from AMD or similar solutions from other companies. I also find it weird that AMD would essentially just concede the high end GPU market for the next 6-8 months until Vega. With a dual GPU part for $600 that might end up slightly faster than 1080 when scaling right, I could see that doing some decent sales damage to Nvidia. That being said, I think Pascal might have some hidden SLI improvements in it. Part of doing NVLink should have an effect on the way they do multi-GPU communication, similar to XDMA (I realize NVLink is server only, but it obviously effects more than just the mezzanine connector). I really wouldn't be surprised if we got a GTX1090, dual GP104.
Meh I've always ignored multi-gpu in favor of single gpu, but yeah nvlink is proof nvidia are rethinking inter-gpu communication so it will invariably trickle down to mgpu at some point
data/avatar/default/avatar10.webp
155.2 FPS vs 215.7 FPS 121.8 FPS vs 182.9 FPS 528.3 FPS vs 749.7 FPS how is that close? am i missing something here or did everyone check the onscreen performance which is capped by vsync at 60fps and ignored offcreen the one that actually matters? :3eyes:
Exactly. This is nothing anywhere near a 390. Is it not really Polaris and just another renumbered card? This whole headline saying it close to 390x is very misleading.
data/avatar/default/avatar34.webp
If those are the results of some incarnation of Polaris 10, then it better just sip at the power draw from time to time since the results are on par or worse than the R9 380 results from the same site: https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=28949748&os1=Windows&api1=gl&hwtype1=dGPU&hwname1=AMD+67DF%3AC4&D2=AMD+ASUS+R9+380+Series EDIT: There needs to be a date of when the tests were run and with what driver version
That is interesting. Even a 380 is beating this 480. Something definitely is fishy here. I hope Hilbert comes and explains exactly why he thinks this card is comparable to a 390X when the numbers say otherwise. Maybe I'm missing something here.
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
That is interesting. Even a 380 is beating this 480. Something definitely is fishy here. I hope Hilbert comes and explains exactly why he thinks this card is comparable to a 390X when the numbers say otherwise. Maybe I'm missing something here.
He doesn't think that based on this numbers, the rumor is its 390x performance and that's in line with the leaked specs. These benchmarks mean exactly ****all because a) it's opengl b) early drivers c) presumably this is Polaris 10 and not 11
data/avatar/default/avatar22.webp
He doesn't think that based on this numbers, the rumor is its 390x performance and that's in line with the leaked specs. These benchmarks mean exactly ****all because a) it's opengl b) early drivers c) presumably this is Polaris 10 and not 11
No, I'm sorry. But I disagree. He clearly states in this article that the numbers from the benchmarks show it performing close the 390x. I agree that these benchmarks don't mean squat. But Hilbert apparently thinks they do since he posted them and specifically made the following statement in the article. "Some performance numbers spotted at an GFXBench run make a rather stong and compelling arguement for the R9 480 (non-X) as it is close to Radeon R9 390X performance"
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
No, I'm sorry. But I disagree. He clearly states in this article that the numbers from the benchmarks show it performing close the 390x. I agree that these benchmarks don't mean squat. But Hilbert apparently thinks they do since he posted them and specifically made the following statement in the article. "Some performance numbers spotted at an GFXBench run make a rather stong and compelling arguement for the R9 480 (non-X) as it is close to Radeon R9 390X performance"
Premature drivers, therefore lackluster OGL performance + possibly ES, lower clocks etc + it's ****ing gfxbench http://i.imgur.com/uiLiDoy.png
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
And people should get it into their heads that's polaris 10 is a TONGA successor with Hawaii-level performance. Jesus christ .
https://forums.guru3d.com/data/avatars/m/225/225084.jpg
Is a 480 going to be a rival for the best selling series of all time the X70? Not sure if a possible 1070 will be as popular as the near 50%(steam stat) 970 is now. I wouldn't even contemplate a 480/480x until all the cards are on the table. Lets be honest the 970 with all it's well documented issues still smashed the field for Nvidia and if pascal is good and 1070 is reasonably priced it will be interesting on which card becomes the next fans favorite and the No1 seller. I'm a fan of video games so i'll most likely always go for the best bang for buck card and not a 2% card like a 980 or 980Ti. If that be an AMD this time around i might get one but i think i can just about get away with my current G1 if i stick to 1080p. So i might skip this gen and wait for Volta or sell 980 and add a few hundred squid, should get me a high end card at least.
data/avatar/default/avatar39.webp
What's the point of buying an AMD graphics card if the drivers are all messed up?
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
Is a 480 going to be a rival for the best selling series of all time the X70? Not sure if a possible 1070 will be as popular as the near 50%(steam stat) 970 is now. I wouldn't even contemplate a 480/480x until all the cards are on the table. Lets be honest the 970 with all it's well documented issues still smashed the field for Nvidia and if pascal is good and 1070 is reasonably priced it will be interesting on which card becomes the next fans favorite and the No1 seller. I'm a fan of video games so i'll most likely always go for the best bang for buck card and not a 2% card like a 980 or 980Ti. If that be an AMD this time around i might get one but i think i can just about get away with my current G1 if i stick to 1080p. So i might skip this gen and wait for Volta or sell 980 and add a few hundred squid, should get me a high end card at least.
I think you'll be fine at 1080p honestly
https://forums.guru3d.com/data/avatars/m/259/259654.jpg
We don't know two things: a) The specs of the systems. b) Which ACTUAL Polaris card this is. That could range from a cut down version of Polaris 11, up to the top model Polaris 10. c) The condition of the Polaris OpenGL driver.
data/avatar/default/avatar20.webp
one word to describe that AMD 67DF : C4 model..."weak"
https://forums.guru3d.com/data/avatars/m/259/259654.jpg
one word to describe that AMD 67DF : C4 model..."weak"
Unless it's the one they put vs the GTX 950 in January. All things need context in this life.
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
Unless it's the one they put vs the GTX 950 in January. All things need context in this life.
the context you're missing is that there are 3 polaris resutls on gfxbench This is the strongest of the 3 if the other two are polaris 11, this is probably the one in the PS4K
data/avatar/default/avatar03.webp
Unless it's the one they put vs the GTX 950 in January. All things need context in this life.
I agree, the reason I said that above is I can't justify the reason it should be "R9 480" and 390(X) equivalent, I call it pure non-sense and a sidegrade not upgrade for a 380X owner, who would now sell a 380 to buy a 380X anyway? πŸ™‚
https://forums.guru3d.com/data/avatars/m/169/169957.jpg
We can get a rough estimate of clocks from the ALU test I posted above by the way, assuming optimization for Polaris is similar to 390x in this test, 390x gets 1000 in this, nice round number polaris gets 70% of that ends up at around 900mhz assuming 2304 SPs i doubt it's polaris 11 because this ***very rough estimate*** would put it at 1600mhz funnily enough it's close to the leaked ps4k clocks, i'd expect it to be clocked higher for PC