|
Excellent. Thank you. As soon as im off work I will do some OCing.
|
#
?
Aug 24, 2020 19:09
|
|
|
|
| # ? May 20, 2024 00:47 |
|
|
ufarn posted:Linus is about to promote the 1usmus custom Ryzen power plan Seems similar to the MSI afterburner auto "OC scanner" I still haven't managed to overclock my ram beyond D.O.C.P, but maybe the latest bios that just dropped will help, I feel like it would work if I bumped the voltage, but I have a million options on my overpriced X570 crosshair VIII, me being 2x16gb and technically not on the qvl doesn't help. Quaint Quail Quilt fucked around with this message at 21:40 on Aug 24, 2020 |
|
#
?
Aug 24, 2020 21:29
|
|
|
I just felt emboldened to try the custom RAM timings, and I can confirm it was still a pain to set up, but I'm at least getting the hang of the SAFE preset without BIOS yelling at me.
ufarn fucked around with this message at 00:00 on Aug 25, 2020 |
|
#
?
Aug 24, 2020 23:52
|
|
|
ufarn posted:I just felt emboldened to try the custom RAM timings, and I can confirm it was still a pain to set up, but I'm at least getting the hang of the SAFE preset without BIOS yelling at me. Overclocking RAM has to be one of the most timing consuming, tedious endeavors that exists, my brain is broken and needs that last few percent of performance for some reason.
|
|
#
?
Aug 25, 2020 00:20
|
|
|
Mr.Radar posted:The Linus Tech Tips video is out, and it's not just a custom power plan but a whole new auto-OC/auto-undervolt tool from 1usmus: If only KINGPIN hadn't swiped the name "Roboclocker" already, that's basically what this is.
|
|
#
?
Aug 25, 2020 02:36
|
|
|
Does ryzen master still not work if you have hyperV / WSL / windows sandbox turned on? That would be a moderately big downside to that program, though not 1usmus's fault. Cool utility though, it's amazing that the software-controlled hardware internals has some far enough that you can now make an automated OC program that's actually intelligent.
|
|
#
?
Aug 25, 2020 02:59
|
|
|
Klyith posted:Does ryzen master still not work if you have hyperV / WSL / windows sandbox turned on? That would be a moderately big downside to that program, though not 1usmus's fault. They have not fixed it as far as I know. That was the main reason I can't use it
|
|
#
?
Aug 25, 2020 03:18
|
|
|
B-Mac posted:Overclocking RAM has to be one of the most timing consuming, tedious endeavors that exists, my brain is broken and needs that last few percent of performance for some reason. It gets even dumber with boards that don't do auto-fallback like my B450 Mortar Ti. It would be great if we can OC anything and do basic stability tests on the fly without the tedious rebooting within the BIOS itself, but I'm not going to get my hopes up from vendors sucking at BIOS design since forever.
|
|
#
?
Aug 25, 2020 04:29
|
|
|
Craptacular! posted:And you would use this over other itx boards, because...? Onboard 2x10Gb NICs. One of my friends is actually torn because he wants mini-ITX for his next desktop, but he also wants a 10G link to his storage array.
|
|
#
?
Aug 25, 2020 05:53
|
|
|
I swear Linus looks just a little more haggard in each thumbnail I see posted ITT
|
|
#
?
Aug 25, 2020 06:21
|
|
|
Munkeymon posted:I swear Linus looks just a little more haggard in each thumbnail I see posted ITT we're in-between major releases so there's not that much grist in the content mill ATM hell even Gamers Nexus is just doing the rounds of making GBS threads on FX processors (not that he's wrong!) and talking about how to properly mount AIOs (which is genuinely useful)
|
|
#
?
Aug 25, 2020 06:48
|
|
|
Klyith posted:Does ryzen master still not work if you have hyperV / WSL / windows sandbox turned on? That would be a moderately big downside to that program, though not 1usmus's fault. Turn off hyper v, turn on ryzen master, run 1usmus's thing to get timings, then go back to your original set up?
|
|
#
?
Aug 25, 2020 08:48
|
|
|
B-Mac posted:Overclocking RAM has to be one of the most timing consuming, tedious endeavors that exists, my brain is broken and needs that last few percent of performance for some reason. So what if you're currently using two sticks of Hynix A-die RAM along with two sticks of Samsung B-die? Both apparently have the same exact base timings, but I haven't bothered trying to tighten them up for fear of something going cattywampus due to some intrinsic differences between the two.
|
|
#
?
Aug 25, 2020 11:35
|
|
|
90s Solo Cup posted:So what if you're currently using two sticks of Hynix A-die RAM along with two sticks of Samsung B-die? Both apparently have the same exact base timings, but I haven't bothered trying to tighten them up for fear of something going cattywampus due to some intrinsic differences between the two. You're not gonna damage anything unless you feed them too much extra voltage or do things with the CPU SoC power. Otherwise the worst thing that happens is you have to clear the CMOS, then keep going. As for what settings to use, I'd use the dram calculator set as if you had 4 sticks of hynix A. Start by writing down all the safe values and trying those, if they work you can try going for fast. The thing I'm wondering about for this situation is: if you have 2 sticks of good ram and 2 sticks of not good ram, and a board with daisy chain topology, is it better for the good ram to be in the close or far slots? I'd guess close, but can think of a decent reason for either answer to be true. (Good close because it can deal with worse signal and reflections in the bad slot, good far because signal is equally bad in both slots with 4 sticks and the bad memory has less delay close.) Palladium posted:It gets even dumber with boards that don't do auto-fallback like my B450 Mortar Ti. Auto-fallback doesn't work if the values are wrong enough to make it hardlock as soon as it sets timings. It's nice but it only works when you're doing a very incremental, one setting at a time approach. But here's my hot tip for memory OCing: when I was doing my memory I crafted a pin jumper on a stick, so I could clear the cmos without pulling the GPU or needing to do the whole thing with the PC opened up on a table. I have some not-ideal-for-ryzen memory and a 1600X -- got it before the ryzem ram problem was common knowledge -- so I needed to go full manual to get good use from it. This involved quite a few cmos clears.
|
|
#
?
Aug 25, 2020 15:53
|
|
|
Klyith posted:You're not gonna damage anything unless you feed them too much extra voltage or do things with the CPU SoC power. Otherwise the worst thing that happens is you have to clear the CMOS, then keep going. Yikes. Here's my tip: If you're going to be in a situation where you know you're going to be clearing CMOS a bunch, and you don't have the older "bridge 1 and 2 for normal operation, bridge 2 and 3 for clear CMOS", wire your case's reset switch into clear CMOS.
|
|
#
?
Aug 25, 2020 16:12
|
|
|
or just get one of these or whatever https://www.amazon.com/Electop-Power-Button-Computer-Switch/dp/B01LMZZFWO
|
|
#
?
Aug 25, 2020 17:03
|
|
|
SwissArmyDruid posted:Yikes. Here's my tip: If you're going to be in a situation where you know you're going to be clearing CMOS a bunch, and you don't have the older "bridge 1 and 2 for normal operation, bridge 2 and 3 for clear CMOS", wire your case's reset switch into clear CMOS. That's a cool idea but would still require pulling the case onto a table and semi-disassembling a bunch of stuff to rewire it. I really didn't want to do more than open the side panel if I didn't need to, because I was doing this long after the initial build. Good on a fresh setup where you're still testing things and haven't gotten it all buttoned up and finished though.
|
|
#
?
Aug 25, 2020 17:17
|
|
|
Klyith posted:That's a cool idea but would still require pulling the case onto a table and semi-disassembling a bunch of stuff to rewire it. I really didn't want to do more than open the side panel if I didn't need to, because I was doing this long after the initial build. wire your CMOS clear switch during build and tape it in an empty optical drive bay (I forget the size of the bay and too lazy to look it up).
|
|
#
?
Aug 25, 2020 18:25
|
|
|
Paul MaudDib posted:Check out the "Asus PN50" which is launched in most markets that aren't the US, it is basically GBP 280 to 380 for the same tier but with 4800u instead of 3800U/etc, and hopefully that translates to $280-380 after removing conversion/tariffs. I think the chassis looks chintzy on the PN50 but the specs look great. Gonna be my main PC for most things besides gaming and uses so little power that I think it will be useful for a long time. Like my Odroid HC2 that is the smallest always-on torrent/NZB machine I could find that holds a 3.5" drive.
|
|
#
?
Aug 26, 2020 06:54
|
|
|
quote:Supercomputer Mahti is now available!
|
|
#
?
Aug 26, 2020 14:09
|
|
|
https://twitter.com/chiakokhua/status/1298433876058988544 This isn't specifically AMD news, but given TSMC are confident enough to take orders for 2nm, it points to what we can expect in a few years. Plus there's only one smaller number after 2nm, then that's it, we'll have the densest chips possible.
|
|
#
?
Aug 26, 2020 22:11
|
|
|
More TSMC stuff: https://twitter.com/aschilling/status/1298223960514080768 3rd slide says N5 Q2 '21, N3 Q1 '23. https://twitter.com/aschilling/status/1298538656362311680 distinct high power vs high density processes on the same node. https://twitter.com/Underfox3/status/1298347518905004033 N5 already has a similar defect rate to N7.
|
|
#
?
Aug 26, 2020 22:20
|
|
|
ConanTheLibrarian posted:https://twitter.com/chiakokhua/status/1298433876058988544 we'll just make a new smaller qualifier but until then 0.01nm go! Though I'm not sure there's going to be much in gains when you make it under 0.1nm since that's the width of an atom and I don't think we're changing that.
|
|
#
?
Aug 26, 2020 22:30
|
|
|
Node names have no relation to actual dimensions.
|
|
#
?
Aug 26, 2020 22:35
|
|
|
gary oldmans diary posted:I've been looking at this intently for a while and it's getting closer. Intel NUCs haven't used CPUs with their best available integrated graphics for 2 generations now and the 4800u version of this is going to really drive that home. The part I'm searching to look for early sellers is Pn50-bbr033md. Provantage has them up for special order, $603. Also I've seen Newegg with the 4300U variant for $329 but it sold out. Bit stiff, the prices I was expecting were more like $500 for the 4800U variant based on Amazon UK listing them at GBP 499, but 4800Us seem to be in incredibly short supply in general so that probably is affecting the prices too. Maybe it'll come down as time goes.
|
|
#
?
Aug 26, 2020 23:20
|
|
|
I'm apprehensive about that site and how long that shipping might take, but the 4300u variant appearing for sale 2 weeks earlier than expected on Newegg is a good sign. I almost bought this (4700u laptop) last week for $530, except reviews mention its RAM being soldered in.
|
|
#
?
Aug 26, 2020 23:32
|
|
|
Provantage is legit, just have a very web-1.0 site. Reminds me of the TigerDirect days lol, straight outta 1999. They cater to enterprise customers who really don't care what the website looks like.
|
|
#
?
Aug 26, 2020 23:40
|
|
|
ConanTheLibrarian posted:https://twitter.com/chiakokhua/status/1298433876058988544 I regret not buying into more TSMC when it was down in the $30s.
|
|
#
?
Aug 27, 2020 02:24
|
|
|
Paul MaudDib posted:Provantage is legit, just have a very web-1.0 site. Reminds me of the TigerDirect days lol, straight outta 1999. Is that where TigerDirect ended up going in terms of focus, too? I remember kind of preferring them to NewEgg for consumer gear but for at least the last 10 years they seem overpriced and their website appears to have been stuck in time, feature-wise.
|
|
#
?
Aug 27, 2020 02:30
|
|
|
movax posted:I regret not buying into more TSMC when it was down in the $30s. Since this is the AMD thread, I'll point out that in the same time period you would have made more money from buying shares of AMD.. but yeah, I hold TSM too, and I'm kinda waiting for the market to react to the recent announcements. Surely the world's leading foundry is worth more than the ~$80 share price it's trading at. I guess the news about China hiring away some TSMC engineers has affected confidence; TSMC is known for having extremely loyal engineers with long tenures at the company. I really can't wrap my head around the motivation for the TSMC guys that left - on the upside, yeah, you are probably getting paid a lot more money than you were at TSMC. Downside is you now work for the CCP and they are investing an unfathomable amount of money with national security implications based on the expectation that you and a handful of other random engineers will be able to huddle up and spitball together a state of the art semiconductor foundry. edit: imagine being a poached TSMC employee showing up on your first day and finding yourself in a room full of washed out former Intel engineers arguing about the best path to 10nm. GRECOROMANGRABASS fucked around with this message at 16:16 on Aug 27, 2020 |
|
#
?
Aug 27, 2020 16:03
|
|
|
It’s time. My i7-940 won’t cut it anymore and games are making me feel nauseous due to low frames. I sat down to price an AMD build and realized the 4xxx line is due this year? I think? Any idea on release date for desktop chips?
|
|
#
?
Aug 27, 2020 21:22
|
|
|
Afaik they've only committed to launching sometime in 2020. B550 and x570 boards will support zen3 though so if you're in a hurry you could grab a 3600 and drop in an upgrade in a few months as long as you don't mind lighting like $160 on fire (probably can recoup some of that on eBay or whatever).
|
|
#
?
Aug 27, 2020 21:27
|
|
|
AARP LARPer posted:It’s time. My i7-940 won’t cut it anymore and games are making me feel nauseous due to low frames. Nobody knows for sure, but AMDs CEO has confirmed the desktop parts will launch before the end of the year. Probably a good bet that they launch after the server versions, and probably a good bet the server versions launch in September so... soon™?
|
|
#
?
Aug 27, 2020 21:27
|
|
|
Cool. Thanks for the update.
|
|
#
?
Aug 27, 2020 23:13
|
|
|
VorpalFish posted:grab a 3600 and drop in an upgrade in a few months as long as you don't mind lighting like $160 on fire (probably can recoup some of that on eBay or whatever). Or do a build and give it to a friend or relative who needs a spare machine now that school is at home. Or do a more bare-bones build and make it a compute node for citizen science. Or pair a 3700 with a B550 mobo now to get a little extra workload headroom and just enjoy the bejeesus out of it. Skip Ryzen 4000 altogether, then rebuild for DDR5/Ryzen 5000 in 18 months-ish (or see if Intel is on the verge of getting their poo poo together yet). Really, no matter what you do it's gonna be amazing compared to a 12 year old quad-core.
|
|
#
?
Aug 27, 2020 23:27
|
|
|
I don't how rational it is, but I still feel like there might be a Zen 3+ refresh before they go to AM5. I plan to upgrade from my 3600X at some point, but probably for a while until after Zen 3 drops and we see where things go. For the time being the 3600X is handling everything I throw at it quite nicely.
|
|
#
?
Aug 28, 2020 00:02
|
|
|
VorpalFish posted:you could grab a 3600 and drop in an upgrade in a few months as long as you don't mind lighting like $160 on fire (probably can recoup some of that on eBay or whatever). The 3600 just got price increased to $200 yesterday, so it's not quite as cheap to use as kindling as it was. Dunno why, but it happened for all major outlets basically overnight. My theory is that they've switched production over to 4000s already but realized they'd run out of 3600s before they were ready to launch. Either that or Intel completely failing to compete at that $200 zone made them realize they didn't need to discount so heavily, as the 10400 sucks and the 10600KF is apparently a paper product. edit: CaptainSarcastic posted:I don't how rational it is, but I still feel like there might be a Zen 3+ refresh before they go to AM5. I plan to upgrade from my 3600X at some point, but probably for a while until after Zen 3 drops and we see where things go. Could be, it's very much going to depend on the DDR5 timeline to availability. Pretty sure that if this year's chaos and delays have impacted that, AMD will know about it and hold back. Klyith fucked around with this message at 00:10 on Aug 28, 2020 |
|
#
?
Aug 28, 2020 00:07
|
|
|
Cygni posted:Nobody knows for sure, but AMDs CEO has confirmed the desktop parts will launch before the end of the year. Probably a good bet that they launch after the server versions, and probably a good bet the server versions launch in September so... soon™? I'm still stupidly holding out hope for Zen 3 Threadrippers this year, but realistically it's got to be Q1/Q2 next year.
|
|
#
?
Aug 28, 2020 00:18
|
|
|
ConanTheLibrarian posted:https://twitter.com/chiakokhua/status/1298433876058988544 Not true, we can start measuring things by picometers.
|
|
#
?
Aug 28, 2020 00:19
|
|
|
|
| # ? May 20, 2024 00:47 |
|
|
wargames posted:Not true, we can start measuring things by picometers.  Introducing our new 300ℓP process!
|
|
#
?
Aug 28, 2020 00:48
|
|