|
Malcolm XML posted:Lrdimm support hell yeah gonna make me a 1tb workstation monster Malcolm XML posted:I will NEVER download more ram Throw your SSD in the trash!!
|
#
?
Sep 10, 2019 01:13
|
|
|
|
| # ? Jun 3, 2024 11:27 |
|
|
512gb of ram is $2500
|
|
#
?
Sep 10, 2019 01:34
|
|
|
Palladium posted:My 3200CL16 E-die runs perfect on XMP undervolted to 1.25V but will not even POST with the Ryzen DRAM calculator safe presets at 3200MHz at 1.35V, what gives? It makes no goddamn sense. For RAM OCing I found it a bit odd. Turn off XMP. Make sure all settings are at AUTO first, then go slowly and change each one, rebooting in between. Set a high -ish voltage for this process to allow the memory to POST/train successfully. You could set just the main 18-20-20-20 timings and leave everything else on auto if you want to be lazy, thats what I did.
|
|
#
?
Sep 10, 2019 01:37
|
|
|
Lube banjo posted:For RAM OCing I found it a bit odd. Turn off XMP. Make sure all settings are at AUTO first, then go slowly and change each one, rebooting in between. Set a high -ish voltage for this process to allow the memory to POST/train successfully. Well, I just got my 16GBx2 Ballistix Sport LT E-die working at 3600C16-19-16-36 on 1.35V, manually keying everything with help of the calculator.
|
|
#
?
Sep 10, 2019 03:10
|
|
|
Palladium posted:Well, I just got my 16GBx2 Ballistix Sport LT E-die working at 3600C16-19-16-36 on 1.35V, manually keying everything with help of the calculator.
|
|
#
?
Sep 10, 2019 03:31
|
|
|
Drakhoran posted:so Gamers Nexus have apparently gotten some info on the upcoming Threadripper platform:
|
|
#
?
Sep 10, 2019 05:33
|
|
|
Anime Schoolgirl posted:is there something wrong with me that i'm somehow seeing loss.jpg in this image I don't see it, can you draw some boxes
|
|
#
?
Sep 10, 2019 05:40
|
|
|
After having my B450M Mortar for a week, I find it baffling to still see ASMedia USB ports making GBS threads themselves with cheapo USB Wi-Fi adapters (random disconnects galore) where Intel USB ports would run them flawlessly. Had the same issue with the Asmedia USB ports on my old 4790K mobo.
|
|
#
?
Sep 10, 2019 07:23
|
|
|
Palladium posted:My 3200CL16 E-die runs perfect on XMP undervolted to 1.25V but will not even POST with the Ryzen DRAM calculator safe presets at 3200MHz at 1.35V, what gives? It makes no goddamn sense. I'm on the 3600 Safe timings on my e-die but it wouldn't post with any kind of tweaked timings at any speed without something like 1.44 for DRAM voltage. Have you set or tried the alternate ProcODT setting as well? That's another thing that stopped me from posting.
|
|
#
?
Sep 10, 2019 09:49
|
|
|
Palladium posted:My 3200CL16 E-die runs perfect on XMP undervolted to 1.25V but will not even POST with the Ryzen DRAM calculator safe presets at 3200MHz at 1.35V, what gives? It makes no goddamn sense. Is it really worth going through this headache for margin of error performance gains in synthetic benchmarks? It's not like you're missing out on a free 1ghz+ overclock on a Sandy Bridge CPU or anything.
|
|
#
?
Sep 10, 2019 12:15
|
|
|
FallenGod posted:Is it really worth going through this headache for margin of error performance gains in synthetic benchmarks? It's not like you're missing out on a free 1ghz+ overclock on a Sandy Bridge CPU or anything. There's virtually no CPU headroom in 3rd gen Ryzens so DDR4 OCing is the only thing left to raise performance. There are nice gains to be had in games from 3200C16 to 3600C16, not just only in synthetics. Anyway I got 3600C16 working fine now, and got an Apple USB-C to 3.5mm adaptor (works with Win10 PCs) to replace the awful onboard audio...So all that's left to fix is Wi-Fi.
|
|
#
?
Sep 10, 2019 12:27
|
|
|
Palladium posted:Apple USB-C to 3.5mm adaptor (works with Win10 PCs) to replace the awful onboard audio I shouldn't be surprised that Apple did the mini displayport cable thing again, but.
|
|
#
?
Sep 10, 2019 14:09
|
|
|
Palladium posted:My 3200CL16 E-die runs perfect on XMP undervolted to 1.25V but will not even POST with the Ryzen DRAM calculator safe presets at 3200MHz at 1.35V, what gives? It makes no goddamn sense. if this is something built into the mobo, maybe it's messing with the subtimings buried on a second page somewhere?
|
|
#
?
Sep 10, 2019 16:57
|
|
|
It's the same thing with my G.Skill 3200 CL14. It's so weird that even the motherboard BIOS's own safe preset doesn't boot. It looked like they used the exact same RAM in GN's recent benchmark so I have no idea what's going on.
|
|
#
?
Sep 10, 2019 17:08
|
|
|
My dual-rank B-die (G.Skill TridentZ 3200CL14, timings 14-14-14-34, 2x16 GB modules) will work at 3600CL14 (timings 14-15-15-30) when I give it 1.45 V. I sometimes have errors in Prime95 but I'm pretty happy with it. The reason I'm posting this: I've noticed that newer versions of the Ryzen DRAM calculator undershoot the voltage needed for normal RAM operation. Version 1.6.0.3 seemed to give better values. By B-die won't even boot at anything less that 1.43 V when operating at 3600CL14. The DRAM calculator suggests 1.37 V  Disclaimer: don't know anything about E-die. You probably don't wanna go to 1.45 V with it.
|
|
#
?
Sep 10, 2019 17:47
|
|
|
Backyarr posted:My dual-rank B-die (G.Skill TridentZ 3200CL14, timings 14-14-14-34, 2x16 GB modules) will work at 3600CL14 (timings 14-15-15-30) when I give it 1.45 V. I sometimes have errors in Prime95 but I'm pretty happy with it. Are you sure all the settings are the same other than the voltage? I don't know what settings require a lot of voltage, but I went from 3600 "Safe" to 3800 "Fast" which actually required less voltage and had some looser timings, but some significantly tighter sub timings. 3600 "Safe" required something like 1.46v but 3800 "Fast" required 1.42.
|
|
#
?
Sep 10, 2019 18:29
|
|
|
Burno posted:Are you sure all the settings are the same other than the voltage? I don't know what settings require a lot of voltage, but I went from 3600 "Safe" to 3800 "Fast" which actually required less voltage and had some looser timings, but some significantly tighter sub timings. 3600 "Safe" required something like 1.46v but 3800 "Fast" required 1.42. They're the same other than the CAD Bus resistances, which I haven't played around with and left at 24-24-24-24. Notice the difference in voltages. Maybe that's the important factor for lower voltages? 1.6.0.3: https://imgur.com/DFwWPGP.png 1.6.2: https://imgur.com/e7uT7yP.png
|
|
#
?
Sep 10, 2019 19:31
|
|
|
Backyarr posted:.... From what I recall, Buildzoid found E-die really likes (works) higher voltage. https://www.youtube.com/watch?v=mhp8iwHJxyI&t=2995s And is indifferent to hair dryers in case you're wondering. https://www.youtube.com/watch?v=OeHEtULQg3Q&t=368s Golluk fucked around with this message at 04:06 on Sep 11, 2019 |
|
#
?
Sep 11, 2019 03:53
|
|
|
Has AMD said anything about when this month the 3950x is supposed to be released?
|
|
#
?
Sep 11, 2019 05:52
|
|
|
Still just "September". I wouldn't expect the launch until late September, possibly even September 30. Since they are still struggling to meet demand for the 3900X they'll want to build up as much stock as possible before launch.
|
|
#
?
Sep 11, 2019 13:48
|
|
|
can't help but wonder if the late 3950X launch and 3900X shortages are a result of AMD simply not getting enough good silicon to meet demand, shamino (ex-Asus employee) hinted that this is also the reason for boost clocks falling short of the official spec. That wouldn't be surprising considering early Ryzen 1 CPUs overclocked much worse than the ones built later as the process matured, what is new is that the CPUs are technically not meeting spec, have virtually no OC headroom and silicon quality now almost equals higher frequency.
|
|
#
?
Sep 11, 2019 14:04
|
|
|
....you know, there was a rumor floating around that Microsoft was tinkering around with AMD-based Surface devices, and there's a Surface event next month. I am almost terrified to hope.
|
|
#
?
Sep 11, 2019 18:07
|
|
|
eames posted:can't help but wonder if the late 3950X launch and 3900X shortages are a result of AMD simply not getting enough good silicon to meet demand, shamino (ex-Asus employee) hinted that this is also the reason for boost clocks falling short of the official spec. probably correct. as process nodes get denser freq becomes really hard to increase. I think they could find fully enabled 8 core dies but those would go to EPYC for margin reasons. So they screwed themselves by promising high frequency cores. AMD marketing creates problems again!
|
|
#
?
Sep 11, 2019 18:26
|
|
|
We're all hoping laptops make the direct jump to ARM, but using AMD to fill the gap probably isn't a terrible idea.
|
|
#
?
Sep 11, 2019 22:27
|
|
|
ufarn posted:We're all hoping laptops make the direct jump to ARM They're called tablets
|
|
#
?
Sep 11, 2019 23:25
|
|
|
Setzer Gabbiani posted:They're called tablets Chromebooks are on both ARM and x86, unless they dropped the ARM ones. not that we are getting off x86 purely for compatibility reasons for a very very long time. There's going to be a middle step of a co-processor before a full switch. There is way too much legacy.
|
|
#
?
Sep 11, 2019 23:29
|
|
|
Setzer Gabbiani posted:They're called tablets Mobile OS though  Not that Windows on ARM is that great right now but linux on ARM isn't bad, it's improved tremendously in the past few years compared to when I first tried it.
|
|
#
?
Sep 11, 2019 23:50
|
|
|
pixaal posted:Chromebooks are on both ARM and x86, unless they dropped the ARM ones. What's the benefit of moving off x86 and onto ARM? All the pundits keep hooting and hollering about moving to ARM. No one ever offers a cogent explanation for why we should.
|
|
#
?
Sep 12, 2019 00:08
|
|
|
Pittsburgh Fentanyl Cloud posted:What's the benefit of moving off x86 and onto ARM? I have no idea, but if it was going to happen there is a 0% chance it would be a clean cut there would be a co processor first, pointing this out tends to get them to stop.
|
|
#
?
Sep 12, 2019 00:10
|
|
|
The ISA is prettier.
|
|
#
?
Sep 12, 2019 00:10
|
|
|
Not sure if I can sell a hardware migration with "but the ISA is prettier"
|
|
#
?
Sep 12, 2019 00:11
|
|
|
People look at the efficiency of low performance ARM CPUs and fancifully imagine that if they scaled up to perform as well as x86 CPUs, they wouldn't be facing the same performance/efficiency tradeoffs that x86 CPUs are.
|
|
#
?
Sep 12, 2019 00:14
|
|
|
|
|
#
?
Sep 12, 2019 00:20
|
|
|
|
|
#
?
Sep 12, 2019 00:24
|
|
|
Pittsburgh Fentanyl Cloud posted:Not sure if I can sell a hardware migration with "but the ISA is prettier" you can if you throw in some leds!
|
|
#
?
Sep 12, 2019 00:44
|
|
|
Lmao I love this chart
|
|
#
?
Sep 12, 2019 00:45
|
|
|
priznat posted:Lmao I love this chart Disco record sales.jpg
|
|
#
?
Sep 12, 2019 00:50
|
|
|
Paul MaudDib posted:Disco record sales.jpg Haha I was thinking Homer’s pumpkin futures but the disco sales one is bang on, forgot about disco stu Probably because he doesn’t advertise
|
|
#
?
Sep 12, 2019 01:02
|
|
|
K8.0 posted:People look at the efficiency of low performance ARM CPUs and fancifully imagine that if they scaled up to perform as well as x86 CPUs, they wouldn't be facing the same performance/efficiency tradeoffs that x86 CPUs are. It's funny, because Chromebooks are a great example of why this doesn't work, outside of the rather expensive ones (which still only perform like low-midrange x86 Chromebooks at best) you get a terrible experience that works in the same way that using your old laptop from 2010 works. pixaal posted:I have no idea, but if it was going to happen there is a 0% chance it would be a clean cut there would be a co processor first, pointing this out tends to get them to stop. None of the successful architecture changes for existing products relied on co-processor nonsense. Historically, it's been a sign of a failed attempts!
|
|
#
?
Sep 12, 2019 01:04
|
|
|
|
| # ? Jun 3, 2024 11:27 |
|
|
The only actual reason that an ARM transition might happen on PCs/servers/whatever is because ARM the company is very liberal about licensing ARM the ISA so there's a much bigger variety of vendors than the x86 duopoly. That's it. If that somehow proves valuable to the market then the transition will happen, otherwise it won't.
|
|
#
?
Sep 12, 2019 03:41
|
|