|
repiv posted:I'm not surprised the performance is more or less what AMD claimed, but I can't figure out what contortion of logic they used to claim better efficiency. http://arstechnica.com/gadgets/2016/07/amd-rx-480-crossfire-vs-nvidia-gtx-1080-ashes/ "In a word: no. In fact, they're not even faster than a GTX 1070 in many games. To be fair to AMD, though, the company only ever said that two RX 480s were faster than a GTX 1080 in one game, under specific settings."
|
#
?
Jul 5, 2016 01:54
|
|
|
|
| # ? May 30, 2024 13:43 |
|
|
SourKraut posted:I'm hoping someone can still comment on this though, since I'm looking to put a 1070 into a Mac Pro (yes, it runs Boot Camp) and am currently thinking the FE or another blower-type might be preferable. Got my MSI 1070 Aero yesterday. Well done B&H and DHL, ordered Thurs and at my office Monday morning. That's impressive for New York to Western Australia. I've only just installed it in a PC here at work, so can't say much, but at first glace it seems quieter at both idle and load compared to my 670, and the blower seems to keep it under the target temps OK. It's definitely longer than the 670, a bit worried it's not going to fit in my Shuttle. I'm going to let it loop on some benchmarks today to see if it throttles like the FE does after a while. Edit: After around 15 minutes, clock is hovering around 1750-1800 in Heaven. Listed speed is 1721, so looks like its doing OK to me. Fan is definitely more noticeable than the 2/3 fan custom jobs, but it's way less annoying than the 670. Just a whoosh of air instead of a very noticeable fan whine. JBark fucked around with this message at 03:02 on Jul 5, 2016 |
|
#
?
Jul 5, 2016 02:05
|
|
|
Your Loyal Vizier posted:Why are the series 10s bad at bitcoins? Because there's literally no reason to use a $800 GPU when you can buy ASICs of similar price that do 4x the performance at bitcoining. Also ASICs as a thing are hilarious that they've taken off so much due to *coin crazes.
|
|
#
?
Jul 5, 2016 02:08
|
|
|
Hubis posted:http://arstechnica.com/gadgets/2016/07/amd-rx-480-crossfire-vs-nvidia-gtx-1080-ashes/ Conditions like unplugging the 1080 from the motherboard?
|
|
#
?
Jul 5, 2016 02:41
|
|
|
Evil Fluffy posted:Conditions like unplugging the 1080 from the motherboard? Conditions like in the article they weren't able to reproduce AMDs results using as identical of a config as possible 
|
|
#
?
Jul 5, 2016 02:43
|
|
|
Am I alone in buying a 1070 as my first nvidia GPU ever, replacing my XFX 290, and moving far away from AMD? Am I the only one succumbing to the hype? Am I wrong?
|
|
#
?
Jul 5, 2016 02:47
|
|
|
skylined! posted:Am I alone in buying a 1070 as my first nvidia GPU ever, replacing my XFX 290, and moving far away from AMD? Am I the only one succumbing to the hype? Am I wrong? gently caress the hype. Is it the right card for the performance you want at the price you want to pay?
|
|
#
?
Jul 5, 2016 02:50
|
|
|
skylined! posted:Am I alone in buying a 1070 as my first nvidia GPU ever, replacing my XFX 290, and moving far away from AMD? Am I the only one succumbing to the hype? Am I wrong? If the 1070 is in your price range, it's definitely the right purchase for the near future
|
|
#
?
Jul 5, 2016 02:52
|
|
|
If anyone's wondering how much stock for RX480 there is, here's a semi-popular shop in germany: http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/Radeon+RX+Serie/RX+480.html The red numbers is how many have sold. Just the sapphire one sold more than the entire 1080 combined: http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/GeForce+GTX+fuer+Gaming/GTX+1080.html But, there have been quite a bit more 1070s than 480s sold, despite the msi one selling for almost 100e over MSRP: http://www.mindfactory.de/Hardware/Grafikkarten+(VGA)/GeForce+GTX+fuer+Gaming/GTX+1070.html So yeah, 1070 is the next 970.
|
|
#
?
Jul 5, 2016 02:57
|
|
|
skylined! posted:Am I alone in buying a 1070 as my first nvidia GPU ever, replacing my XFX 290, and moving far away from AMD? Am I the only one succumbing to the hype? Am I wrong? You are not alone. 
|
|
#
?
Jul 5, 2016 03:03
|
|
|
Contrary to what AMD said, some RX480 4GB owners are reporting it does have the full 8GB of 8ghz GDDR5 on board and is just BIOS locked down to 4GB at 7ghz. The dream is alive 
|
|
#
?
Jul 5, 2016 03:07
|
|
|
i need a video card by August 4th. Is the 1070 gonna be cheaper by August 4th? all this inside baseball stuff is so extra
|
|
#
?
Jul 5, 2016 03:11
|
|
|
repiv posted:Contrary to what AMD said, some RX480 4GB owners are reporting it does have the full 8GB of 8ghz GDDR5 on board and is just BIOS locked down to 4GB at 7ghz. 
|
|
#
?
Jul 5, 2016 03:44
|
|
|
TenementFunster posted:i need a video card by August 4th. Is the 1070 gonna be cheaper by August 4th? all this inside baseball stuff is so extra If you're in the US, probably not. Availability might be good
|
|
#
?
Jul 5, 2016 03:51
|
|
|
TenementFunster posted:i need a video card by August 4th. Is the 1070 gonna be cheaper by August 4th? all this inside baseball stuff is so extra Best deal seems to be the Gigabyte WindForce for $400 http://www.newegg.com/Product/Product.aspx?Item=N82E16814125875
|
|
#
?
Jul 5, 2016 04:33
|
|
|
Psy890 posted:Best deal seems to be the Gigabyte WindForce for $400 $400 for a AIB 1070 is awesome if 480 AIBs are going for ~$270. Truga posted:If anyone's wondering how much stock for RX480 there is, here's a semi-popular shop in germany: Die-harvested 314mm2 GTX 1070 is 35% larger than the fully enabled 232mm2 RX480, rest of the board costs should be the same since they use 8GB GDDR5 on 256 bit bus, but the 1070 retails for at least 66% more. If Nvidia isn't making >90% of overall discrete GPU profit by this point they are doing something wrong. Palladium fucked around with this message at 04:55 on Jul 5, 2016 |
|
#
?
Jul 5, 2016 04:40
|
|
|
Oh no, it doesn't support DDR4  . How will I download more RAM anymore?
|
|
#
?
Jul 5, 2016 07:24
|
|
|
Hubis posted:http://arstechnica.com/gadgets/2016/07/amd-rx-480-crossfire-vs-nvidia-gtx-1080-ashes/ Am I taking crazy pills? Look at the video of the presentation, starting at around 1:05. "On the right is the 700$ solution, left is something much better." A couple of seconds later, as he says: "They deliver basically the same performance. Maybe people with really crazy eyes can spot millisecond differences", the scene on the left ("something much better") fails to render 80 percent of the game action and the article doesn't even mention that? 
|
|
#
?
Jul 5, 2016 10:12
|
|
|
amd is the one taking crazy pills, if their products were better they would say exactly that instead of using bullshit qualifiers like performance per inch in their promos It absolutely isnt rendering the same thing as the 1080 in many spots. The terrain looks significantly worse for CF 480s for most of it. The 1080 side almost always had as many but usually many more units on screen Fauxtool fucked around with this message at 10:57 on Jul 5, 2016 |
|
#
?
Jul 5, 2016 10:52
|
|
|
mcbexx posted:Am I taking crazy pills? Fauxtool posted:amd is the one taking crazy pills, if their product was better they would say exactly that instead of using bullshit qualifiers like performance per inch AMD made a press release on this saying it's actually the 1080 rendering less. That's even less wiggly room if you want to take them to task for it, but it's and old topic and I'm not interested in dredging on something that seemed pretty conclusively answered. BTW, AMD will apparently be dropping a Vbios update for RX480s to fix the PCIE power draw? I guess the reference board can properly redistribute how power is allocated so there won't be any compromise in clockspeeds, just the 6 pin (which is actually a truncated 8 pin apparently) sucking it's proper share. RIP fewer motherboards, RIP more OEM PSUs? That's not on AMD though, that's on lovely OEMs. For hire: previous AMD vbios coder.
|
|
#
?
Jul 5, 2016 11:07
|
|
|
Fauxtool posted:amd is the one taking crazy pills, if their products were better they would say exactly that instead of using bullshit qualifiers like performance per inch in their promos According to Oxide, the pre-release 1080 driver that AMD used had a bug that caused it to generate less snow than intended. The CF480s terrain looked worse because it was obscured by more plain boring snow. I don't think the different number of units on screen was ever addressed though. Maybe Ashes is just a bad benchmark 
|
|
#
?
Jul 5, 2016 11:17
|
|
|
they should really be benchmarking on overwatch since its what everyone is actually playing and it will be relevant data for what kind of components to buy rx 480 does quite well per dollar on overwatch
|
|
#
?
Jul 5, 2016 11:20
|
|
|
Fauxtool posted:they should really be benchmarking on overwatch since its what everyone is actually playing and it will be relevant data for what kind of components to buy
|
|
#
?
Jul 5, 2016 11:23
|
|
|
Ashes *is* a bad benchmark Part of the game's claim to fame is that poo poo is procedurally generated every time you play it. Last I heard, this also extended to the benchmark as well. And if there's one thing that you absolutely don't want when benchmarking things, it's a random seed changing itself from run to run. Even if the seed is exactly identical every time the benchmark is run, we're still just supposed to take their word that the result should be the same, regardless of hardware? Please. That said, it is in fact true that the Nvidia card there was not executing shaders properly, but it doesn't change the fact that Ashes is a lovely benchmark.
|
|
#
?
Jul 5, 2016 11:25
|
|
|
Josh Lyman posted:My 3570K integrated graphics would probably do well per dollar on Overwatch it would be if you could actually quantify its value. They would need to make a version without integrated graphics you can price against
|
|
#
?
Jul 5, 2016 11:28
|
|
SwissArmyDruid posted:Ashes *is* a bad benchmark Part of the game's claim to fame is that poo poo is procedurally generated every time you play it. Last I heard, this also extended to the benchmark as well. And if there's one thing that you absolutely don't want when benchmarking things, it's a random seed changing itself from run to run. Even if the seed is exactly identical every time the benchmark is run, we're still just supposed to take their word that the result should be the same, regardless of hardware? Please. Thats what multiple trials are for! Assuming an even distribution of random seeds mapped to scene complexity, statistically speaking you are more likely to get more varied results than just seeds that get you complex scenes every time.
|
|
|
#
?
Jul 5, 2016 12:35
|
|
|
Watermelon Daiquiri posted:Thats what multiple trials are for! Assuming an even distribution of random seeds mapped to scene complexity, statistically speaking you are more likely to get more varied results than just seeds that get you complex scenes every time. ....I don't know about you? But I'm not looking forward to reviewers having to do 100+ runs of a benchmark just to get a statistically significant result just because of a random seed generator.
|
|
#
?
Jul 5, 2016 12:44
|
|
|
Clearly the only true way to benchmark a GPU's performance is to test each GPU of a large set multiple times (n> 30) using a large amount of a random assortment of benchmarking software run by a randomly selected assortment of unbiased reviewers. Only then will we know which card run gam best.
|
|
#
?
Jul 5, 2016 12:48
|
|
|
God, I hope there's a /s in there somewhere.
|
|
#
?
Jul 5, 2016 12:54
|
|
|
For real though ashes sucks and has been skewing bemchmarks for like 6 months straight now. There are other, real, dx12 games now. When AMD put up that slide I legitimately laughed.
|
|
#
?
Jul 5, 2016 12:55
|
|
|
Pssshh all you need is a script to automate the benchmarking.
|
|
|
#
?
Jul 5, 2016 13:02
|
|
|
SwissArmyDruid posted:God, I hope there's a /s in there somewhere. I mean, technically, yes this is the only way to get truly statistically accurate and meaningful test results in these situations. Doing what I said would yield a result that would be near irrefutable. Of course it's infeasible, and people instead opt to just take manufacturers at their word and to trust reviews by people who receive a free graphics card from a manufacturer who run a single benchmark. I would prefer it if a review just either bought or received 5-10 of the same cards and ran them all in the exact same rig running the exact same test, and then doing the same for other benchmarking softwares. A lot of these reviews say things like "but ours has X limitation/boon, so you'll expect it to run better/worse depending on your situation. Also we tested it once." Just have one institute do a test correctly according to accepted modern statistics. jokes fucked around with this message at 13:19 on Jul 5, 2016 |
|
#
?
Jul 5, 2016 13:14
|
|
|
Watermelon Daiquiri posted:Pssshh all you need is a script to automate the benchmarking.
|
|
#
?
Jul 5, 2016 13:17
|
|
|
Josh Lyman posted:Lol if you think tech reviewers have any technical skills I think Phoronix use a script to run their test suite. Of course it would be the Linux-centric review site that does it 
|
|
#
?
Jul 5, 2016 13:23
|
|
|
Would a GTX 1080 with its clock speed reduced by 50% give a reasonable prediction of GTX 1060 performance? Does the 1080 overclocking software allow for substantially decreased clock speed? When is RX 470 coming?
|
|
#
?
Jul 5, 2016 14:02
|
|
|
PBCrunch posted:Would a GTX 1080 with its clock speed reduced by 50% give a reasonable prediction of GTX 1060 performance? Best thing you could do is flash a modded bios that only enables n cuda cores, where n is whatever 1060 has, then set clocks to whatever 1060 runs at. It'll still be a bit different due to gddr5x, but close enough probably.
|
|
#
?
Jul 5, 2016 14:08
|
|
|
So today's the day AMD explains exactly how they hosed the chicken with the 480 pci-e slot. Any bets on what it was or how theyll propose to fix it? Also, why the gently caress have none of the aftermarket card sellers announced actual release dates? Do y'all think the card problems will delay those launches?
|
|
#
?
Jul 5, 2016 14:18
|
|
|
Hieronymous Alloy posted:So today's the day AMD explains exactly how they hosed the chicken with the 480 pci-e slot. Any bets on what it was or how theyll propose to fix it? Maybe AMD hoarded all the chips to stock up on the reference cards 
|
|
#
?
Jul 5, 2016 14:21
|
|
|
Apparently I grabbed Ashes on steam thinking it might be an OK Supreme Commander successor. Boy was I wrong, that game loving sucks. The only relevant benchmark that should be done using it is how long you can stand to play it before the boredom becomes terminal.
|
|
#
?
Jul 5, 2016 14:33
|
|
|
|
| # ? May 30, 2024 13:43 |
|
|
Is it possible that, as Global Foundries improves its finfet technique, the yields become higher quality/allow for higher clocks? I guess the same question goes to TSMC..will the yields improve chip quality later in the GPU's generation?
|
|
#
?
Jul 5, 2016 14:33
|
|