|
Vir posted:When Asus and EVGA even went as far as changing the board design, and EVGA even confirmed that they couldn't reach full spec with just polymer caps, that does suggest that it was a real issue though. If they can work around it in the drivers in a way that doesn't even lower performance, then that's a good save from Nvidia. yeah, they announced it today. at this point there has been enough FUD stirred up that if you don't change your board you're going to lose sales permanently, even if it's not technically necessary. we don't (and will never) know the exact nexus between power supply instability, cap problems, driver problems, boost behavior, generally lower node performance, binning, and clockrates. Did EVGA just set a too-aggressive clockrate and their cards couldn't deliver it? We will never know. But other vendors had no problems with their SP-CAP based cards. I just think the difference in tone between AMD problems and NVIDIA problems is pretty funny. A year after launch AMD cards still randomly blackscreen every 15 minutes: "mAyBe yOuRe hOlDiNg tHe pSu wRoNg!". NVIDIA has a launch problem and fixes it in a week with a driver update without loss of performance: "gently caress YOU NVIDIA GIVE ME MY MONEY BACK". People are just so willing to automatically assume the worst about NVIDIA, 100% of the time, and it's usually baseless. And they are willing to give AMD the benefit of the doubt every time whatever the problem. I didn't see anyone running loops of NOPs to prove that their 3080 was actually stable and NVIDIA didn't technically market the cards falsely - and you bet that if NVIDIA's patch had dinged performance even a percent that people would be clamoring for a class-action suit. Paul MaudDib fucked around with this message at 20:55 on Sep 28, 2020 |
# ? Sep 28, 2020 20:52 |
|
|
# ? Jun 6, 2024 06:37 |
|
Paul MaudDib posted:I just think the difference in tone between AMD problems and NVIDIA problems is pretty funny. A year after launch AMD cards still randomly blackscreen every 15 minutes: "mAyBe yOuRe uSiNg tHe pSu wRoNg!". NVIDIA has a launch problem and fixes it in a week with a driver update without loss of performance: "gently caress YOU NVIDIA GIVE ME MY MONEY BACK". K8.0 posted:Paul this isn't even a good troll by your standards. The AMD CPUs hit their review performance, the Nvidia GPUs may not.
|
# ? Sep 28, 2020 20:55 |
|
oh no paul said something good about NVIDIA, the world is ending like c'mon, everyone knows NVIDIA takes a bunch of completely unfounded poo poo from people who just hate their guts for what amounts to outcompeting AMD and other competitors. Look at the immediate, reflexive poo poo the entire tech sector took on them for the ARM acquisition. Cause Jensen is going to buy a $40b IP built entirely on licensing and then burn it to the ground by refusing to license it to anyone. they are like Apple, it's trendy to hate on NVIDIA. That certainly played a role here. The difference in the reaction of the community between the Navi problems and this is just night and day. People pulled out every stop to defend AMD and insist that it must have been user error. A week after launch and people are clamoring for NVIDIA to recall product and cut them a fat check. Paul MaudDib fucked around with this message at 21:05 on Sep 28, 2020 |
# ? Sep 28, 2020 20:59 |
|
Well my free 12-pin Seasonic cable arrived at least. I filled out the forum on Thursday after snagging a 3090, they shipped from Taiwan the next day and it arrived late this morning.
|
# ? Sep 28, 2020 21:00 |
|
9/23 gang got an update! No shipping email, but reference number search on fedex finally showed up. Only a shipping label created, but estimated delivery of 10/2.
|
# ? Sep 28, 2020 21:07 |
|
Paul MaudDib posted:oh no paul said something good about NVIDIA, the world is ending I mean when you present yourself as the premium brand and charge for products at premium prices, yeah expectations are higher. Totally agree this got blown out of proportion but it comes with territory. Goes the same for any other product. The higher end vendor with the premium takes more flak over smaller issues. But it works both ways too, the lol AMD driver memes even though they are pretty much fixed still dog AMD months after resolution. No one believes that they can deliver a top notch GPU either and they are surely losing sales because no one is bothering with Big Navi.
|
# ? Sep 28, 2020 21:08 |
|
Posting again, as it relates to previous posts:exquisite tea posted:Is tomorrow supposed to be the day Amazon starts to stock 3080s, or Tuesday? Posted yesterday, and: Sagebrush posted:The only place that can "handle it" at this level of demand and scalping is Amazon, and they appear to have not yet received a single shipment of cards. Any idea when Amazon will start selling the 3080?
|
# ? Sep 28, 2020 21:08 |
|
Samadhi posted:Any idea when Amazon will start selling the 3080?
|
# ? Sep 28, 2020 21:12 |
|
Dilber posted:Am i first? Someone photoshop the guy from Titanic falling into that giant boat propeller.
|
# ? Sep 28, 2020 21:17 |
|
Pretty sure that card is the entire size of the XB Series S.
|
# ? Sep 28, 2020 21:25 |
|
Kingnothing posted:9/23 gang got an update! 9/23 gang rise up, I just got a shipped email a few minutes ago. Scheduled for tomorrow before noon, part of me hopes it'll get pushed to Wednesday cause I'm not home in the morning
|
# ? Sep 28, 2020 21:27 |
|
AMD absolutely does not get credit for having their drivers 'pretty much fixed' by the time of the entirely next generation launch.
|
# ? Sep 28, 2020 21:28 |
|
So what's a realistic expectation for TSMC instead of Samsung? Can they even replace them by the Super refresh, or are we talking RTX 4000?
|
# ? Sep 28, 2020 21:29 |
|
Bezos you useless gently caress
|
# ? Sep 28, 2020 21:44 |
|
Alchenar posted:AMD absolutely does not get credit for having their drivers 'pretty much fixed' by the time of the entirely next generation launch. Sure they do. It's called Finewine and it's been a favorite topic of discussion among techtubers for years. Who doesn't love seen how the R9 380 caught up to the GTX 970 2 years after both cards were EOL'd.
|
# ? Sep 28, 2020 21:48 |
|
My Gigabyte 3080 arrived about an hour ago and I just finished running benchmarks for the first time so I can compare results to Air Raid and the dentist. It has been a long road getting here and I have to give a special shoutout to a (liquidation?) company in Missouri called sssence solutions who hooked me up with a modular PCI-e cable for my PSU. It uses these stupid rockwell type connectors and apparently I got rid of all of my extras thinking "I'll never need more than two 8-pins," but they sold me an extra they somehow had laying around. They're absolutely impossible to find beyond buying a used $100+ PSU just for the cable, in which case it obviously makes more sense to just get a new PSU, so I'm appreciative of their help. The big downside is that the PSU is semi-modular and the two 8-pins I had were split from one cable in the big bunch, so I now have three 8-pins, one of which is unconnected and dangling under the card. It's ugly as hell and I will need to come up with some solution that doesn't partially block airflow into one of the fans.code:
Gigabyte Z170X-Gaming 7 rev 1.0 i7 6700k boost to 4.5 GHz 16GB G.Skill F4-3400C16D-16GTZ SATA3 SSD I forgot to save all but the first test locally, but I did save the first Port Royal benchmark results. GPU clocks topped out at 1950 MHz for about 4 seconds and then hovered around 1890-1910 MHz for most of the rest of the run. All benchmarks run on today's drivers. My only real issue at the moment is that the card seems to be running quite a bit warmer than other Gigabyte 3080 Gaming OC card numbers I've seen. Average temps during the benchmarks are around 71C. I recently replaced my single, dinky 120mm front case fan with two 140mm fans. Replaced the other two 120mm fans with new ones of the same size, as well. The card is so thick that my lower, front fan is blasting air into the front of the card instead underneath. My 3.5" HDDs are hidden underneath the case's power supply shroud and the fans are mounted as low as they can possibly go, so I'm not entirely sure what I can do about it beyond cranking up the GPU fans.
|
# ? Sep 28, 2020 21:50 |
|
Bloodplay it again posted:My only real issue at the moment is that the card seems to be running quite a bit warmer than other Gigabyte 3080 Gaming OC card numbers I've seen. Average temps during the benchmarks are around 71C. That is hotter than mine... it never goes over 66-67C really, with mediocre case airflow. Slightly higher scores too, but then your CPU is way faster than mine. Core clocks are about the same tho, 1950 sometimes, 1920-1935 most of the time. Also, this looks like a steal
|
# ? Sep 28, 2020 21:55 |
|
New Nvidia driver just dropped. Time will tell if it fixes the crashing issues you people are having
|
# ? Sep 28, 2020 22:05 |
|
guess who found his extra power supply cables and prepare for Numbers split cable 456.38: port royal: 1: 11432 2: 11383 3: 11409 4: 11395 5: 11416 6: 11402 7: 11384 8: 11453 9: 11388 10: 11409 avg: 11407 min: 11383 max: 11453 2x pci-e cables 456.38: port royal: 1: 11537 2: 11514 3: 11585 4: 11588 5: 11582 6: 11612 7: 11574 8: 11582 9: 11581 10: 11574 avg: 11573 min: 11514 max: 11612 2x pci-e cables 456.55: port royal: 1: 11572 2: 11561 3: 11558 4: 11536 5: 11614 6: 11560 7: 11540 8: 11547 9: 11560 10: 11559 avg: 11561 min: 11536 max: 11614 so no extra performance on new drivers in this particular scenario now i didn't get pre-driver/cables outside of an old saved test: timespy 456.38, split pci-e cables: 1: 14890 (Graphics 18024, CPU 7501) timespy 456.55, 2x pci-e cables: 1: 14754 (Graphics 17966, CPU 7329) 2: 14717 (Graphics 17958, CPU 7277) 3: 14666 (Graphics 17932, CPU 7218) 4: 14670 (Graphics 17939, CPU 7218) 5: 14629 (Graphics 17921, CPU 7168) just one tiny caveat... i'm still power limited on this 650W anyway... change cpu from 4.7GHz to 3.7GHz on the off chance we can squeeze some extra power: timespy: 1: 13940 (Graphics 17553, CPU 6435) 2: 13893 (Graphics 17557, CPU 6366) 3: 13913 (Graphics 17584, CPU 6374) port royal: 1: 11565 2: 11567 3: 11561 well there goes that theory - although i am not power limited for a couple of frames. core clock will still hit 2010mhz at peak in port royal but generally sticks in the 1950-1965 range. towards the end of the benchmark the gpu will hit 71c but i've adjusted it to a custom curve for a fixed rpm at a low noise level accounting for peak. i could just run everything at 100% and get the thermals ignored if i were chasing score though and yes now i am tempted to grab a 850w and see what changes but that's nothing to do with extra performance and i won't
|
# ? Sep 28, 2020 22:05 |
|
Aren't most of the GPU thermal benchmarks done on open-bench setups because they're all lazy for the most part anyway. I remember Tech Jesus complaining about this from other reviewers
|
# ? Sep 28, 2020 22:11 |
|
My Zotac 3080 hasn't crashed a single time and is boosting over 2K with a slightly more agressive fan curve despite having the BAD CAPS that all the electrical engineers on reddit are up in arms about
|
# ? Sep 28, 2020 22:12 |
|
please post those benchmark numbers in the form of a graph if you expect anyone to read them
|
# ? Sep 28, 2020 22:12 |
|
SaltPig posted:Well my free 12-pin Seasonic cable arrived at least. I filled out the forum on Thursday after snagging a 3090, they shipped from Taiwan the next day and it arrived late this morning. Nice, I have to do this but I gotta pull out the PSU from my case to get at the label they want.
|
# ? Sep 28, 2020 22:12 |
|
Sagebrush posted:please post those benchmark numbers in the form of a graph if you expect anyone to read them also my card still holds stable in gpu-intensive games at 2010MHz post-driver update
|
# ? Sep 28, 2020 22:19 |
|
FYI when you see Power Limit in GPU-Z or equivalent, it's not talking about your PSU. That's your graphics card firmware limiting the power. People here saying "I'm PSU power limited..." but that's not really how it works. If you really max out your PSU, things may start crashing or the PSU will turn off, but you're not going to get the GPU nicely downclocking itself. It doesn't care about any part of the system other than itself.
|
# ? Sep 28, 2020 22:28 |
|
Wiggly Wayne DDS posted:guess who found his extra power supply cables and prepare for Numbers that you're getting that on a 6700K is really good. I admit I'm a little surprised mine is tending to be a bit lower than I'd expect and otherwise see from other people, but it's still good. Just wondering if maybe I have a slightly bad configuration somewhere. My install is a bit poo poo and quite old now, carrying over a lot of garbage. and I also can't run on UEFI mode and need to run on CSM because it won't POST to my M.2 for some reason, but that shouldn't matter Xaris fucked around with this message at 23:06 on Sep 28, 2020 |
# ? Sep 28, 2020 22:54 |
|
Xaris posted:FYI this is almost exactly what I'm getting on Ryzen 3600, 2x16GB Corsair Ballistix (micron e-die, DRAM OC'd using Ryzen DRAM Calc to SAFE settings but I think t here's something a bit wonky), MSI B450 Tomahawk Max, with EVGA 3080 XC3 Gaming (+112/+200 on clock/mem), 650W Corsair RM650 on 2x split PCI-E cables. Time Spy varies a bit but hit 14153 and Port Royale is nominally around 11,301. Have also hit 13,500 TS and 10,251 PR at times too, but I think thats more having other applications running on my second monitor and I had Instant-Replay on when I did that. Latest drivers Ballistix is a Crucial brand btw.
|
# ? Sep 28, 2020 22:58 |
|
8-bit Miniboss posted:Ballistix is a Crucial brand btw. kinda regret the ryzen choice, I know Intel commands an unnesssecary markup but dealing with Intel/mobo/dram is a lot easier than Ryzen and PBO and stuff is really a bit awkward. next build ill go back to intel.
|
# ? Sep 28, 2020 23:06 |
|
Xaris posted:FYI this is almost exactly what I'm getting on Ryzen 3600, 2x16GB Corsair Ballistix (micron e-die, DRAM OC'd using Ryzen DRAM Calc to SAFE settings but I think t here's something a bit wonky), MSI B450 Tomahawk Max, with EVGA 3080 XC3 Gaming (+112/+200 on clock/mem), 650W Corsair RM650 on 2x split PCI-E cables. Time Spy varies a bit but hit 14153 and Port Royale is nominally around 11,301. Have also hit 13,500 TS and 10,251 PR at times too, but I think thats more having other applications running on my second monitor and I had Instant-Replay on when I did that. Latest drivers So what you're saying is your Overclocked card is getting slightly lower numbers than Wiggly Wayne's stock card?
|
# ? Sep 28, 2020 23:09 |
|
Xaris posted:all c-words sound the same Intel really isn't worth it price/perf unless you're chasing some crazy FPS for a high refresh rate monitor in esports titles.
|
# ? Sep 28, 2020 23:11 |
|
Twerkdad posted:9/23 gang rise up, I just got a shipped email a few minutes ago. Scheduled for tomorrow before noon, part of me hopes it'll get pushed to Wednesday cause I'm not home in the morning No shipped email, but someone scanned it onto a truck: It's got 1-day shipping (gottagofast) so hopefully I'll have something by 8PM tomorrow night?
|
# ? Sep 28, 2020 23:12 |
|
AirRaid posted:So what you're saying is your Overclocked card is getting slightly lower numbers than Wiggly Wayne's stock card? They both have OC cards, dont they? The Gigabyte Gaming is factory oc'd .
|
# ? Sep 28, 2020 23:15 |
|
jisforjosh posted:Intel really isn't worth it price/perf unless you're chasing some crazy FPS for a high refresh rate monitor in esports titles. The only thing I really am disappointed about with my 10700k is that I don't seem to have won the silicon lottery. Mind you, I also don't have an AMD platform to do a compairson with, so if anyone wants to send me a B550 mITX board and a 3600 or better chip, I'm down to set up a blind taste test.
|
# ? Sep 28, 2020 23:15 |
|
Does Zen 3 come out on the 8th or is that just when it's getting revealed with no release date? Wondering when I can grab one of those bad boys.
|
# ? Sep 28, 2020 23:21 |
|
jisforjosh posted:Intel really isn't worth it price/perf unless you're chasing some crazy FPS for a high refresh rate monitor in esports titles. I actually just yesterday replaced my old 4x8gb 3000 GSKILLZ ram that, for the very same manufacturing number, had 2 different sets of timings, dies, and ranks, with a single 2x16GB 3200 CL16 ram and that pretty much added +600 right there just swapping that around and a little more doing manual timings. Kinda crazy how sensitive it is to ram even just on a "mostly" GPU Test like Port Royale. Although tbf my Windows install is really a big long-running mess at this point and still can't use UEFI because it wont post to my M.2 drive (probably because it got confused with an old SSD when I installed it) so I'm still on CSM mode--It's just too much work to unplug everything else, wipe, and reinstall all my engineering programs to fix that but I'll probably tackle it when I get some time. Xaris fucked around with this message at 23:24 on Sep 28, 2020 |
# ? Sep 28, 2020 23:22 |
|
Xaris posted:all c-words sound the same It's not that bad. My Ballistix 3600 2x16GB memory worked without any problems on my X570 Tomahawk. It's pretty straightforward and a lot of the troubles date back to Ryzen's debut CPUS in the 1700 range. Having said that I'd go back to Intel the moment they can offer a more efficient, cooler CPU that continues to deliver the best FPS in games.
|
# ? Sep 28, 2020 23:22 |
|
Intel is probably getting hellfucked by Zen 3 lmao I say this as someone who has been on intel forever.
|
# ? Sep 28, 2020 23:28 |
|
joke's aside this was primarily trying to investigate the crashing issue of stock AIB gpus running with GPUBoost to nominally too high frequencies on their ownXaris posted:FYI this is almost exactly what I'm getting on Ryzen 3600, 2x16GB Corsair Ballistix (micron e-die, DRAM OC'd using Ryzen DRAM Calc to SAFE settings but I think t here's something a bit wonky), MSI B450 Tomahawk Max, with EVGA 3080 XC3 Gaming (+112/+200 on clock/mem), 650W Corsair RM650 on 2x split PCI-E cables. Time Spy varies a bit but hit 14153 and Port Royale is nominally around 11,301. Have also hit 13,500 TS and 10,251 PR at times too, but I think thats more having other applications running on my second monitor and I had Instant-Replay on when I did that. Latest drivers Pivo posted:FYI when you see Power Limit in GPU-Z or equivalent, it's not talking about your PSU. That's your graphics card firmware limiting the power. People here saying "I'm PSU power limited..." but that's not really how it works. If you really max out your PSU, things may start crashing or the PSU will turn off, but you're not going to get the GPU nicely downclocking itself. It doesn't care about any part of the system other than itself. i wasn't really expecting to be truly maxing out my psu but that the psu had split limits on where it can distribute power, but i can't find any thorough documentation covering the evga supernova g3 650 other than the combined rating for 12v Symetrique posted:They both have OC cards, dont they? The Gigabyte Gaming is factory oc'd . sean10mm posted:Intel is probably getting hellfucked by Zen 3 lmao
|
# ? Sep 28, 2020 23:33 |
|
sean10mm posted:Intel is probably getting hellfucked by Zen 3 lmao the only thing left for amd to do with ryzen is get the clocks up and once they do that, buh bye intel
|
# ? Sep 28, 2020 23:34 |
|
|
# ? Jun 6, 2024 06:37 |
|
Taima posted:Does Zen 3 come out on the 8th or is that just when it's getting revealed with no release date? They haven't said either way. The running expectation is that on the 8th AMD will officially announce/release Zen3, with actual ability to buy it hitting a week or two later. Kinda like with NVidia did announcing Ampere on the 1st but you couldn't buy them until the 17th. e; also no one knows what sort of availability they'll have. A ton of nerds have been salivating over Zen 3 for months, so it's not unreasonable to expect that it'll be difficult to get some of the more popular chips for a little.
|
# ? Sep 28, 2020 23:36 |