|
Issues are pretty overblown, I haven't checked my temps but I have a B450 Tomahawk running a 3700x with 3600CL16 RAM (thinking about setting what I can get it to run at at CL14 or even CL12, seems to be more important, but that's also something for the future since I'm sure it will change with BIOS updates and this is already more power than I currently need). I had an issue where the CPU error light would come on on the motherboard for a solid 10 seconds before posting. It seems fixed magically there's no new BIOS updates and I haven't noticed it do this in the last 3 or 4 days it just comes right on. It's not sleeping no RGB, I've even pulled power to make sure it wasn't sleeping nearly instant power button to desktop, NVMe is fast as hell. BIOS updating was easy, just follow the directions, you need to rename the BIOS (and change the extension! make sure show extensions is enabled if using windows). You also need it on the correct format drive and it has to be the only partition and it has to take up the entire drive. This means you are limited to under a certain size since you hit partition size limit of FAT. I'm also used to running my CPU hot, my 2500k pretty much idled at 60c and spiked to 85c during gaming for the last 6 years or so. I seriously don't care what my CPU is running at, it's just nice to not have my GPU fan wailing at me from all the heat my 2500k was dumping out. pixaal fucked around with this message at 20:45 on Jul 30, 2019 |
#
?
Jul 30, 2019 20:40
|
|
|
|
| # ? Apr 18, 2024 02:48 |
|
|
https://www.youtube.com/watch?v=nugwAOvijHQ
|
|
#
?
Jul 30, 2019 20:52
|
|
|
https://www.anandtech.com/show/14688/amd-releases-new-chipset-drivers-for-ryzen-3000-more-relaved-cppc2-upscaling New chipset drivers.
|
|
#
?
Jul 30, 2019 23:09
|
|
|
AMD actually lowered the performance of their chips in response to stupid people on Reddit complaining about a nonexistent issue, I mean it might be a totally insignificant difference but I still don't like it.
|
|
#
?
Jul 31, 2019 00:34
|
|
|
That Anandtech article says a system that would previously idle at ~2.2GHz now idles at 3GHz. I don't like this. I'd rather see that poo poo clocking low when my PC is doing nothing. Also, will this be pushed out as part of a FW update or is it something handled by the OS?
|
|
#
?
Jul 31, 2019 06:15
|
|
|
Chipset drivers usually are Windows installers that modify the system’s power management. If you’re in a Linux environment, you won’t install them and you won’t have any difference because systemd does all this instead.
|
|
#
?
Jul 31, 2019 06:35
|
|
|
MaxxBot posted:AMD actually lowered the performance of their chips in response to stupid people on Reddit complaining about a nonexistent issue, I mean it might be a totally insignificant difference but I still don't like it. To be fair, there's something of an actual problem with the high reported idle temperatures causing fans to ramp when they don't really need to. But, that can be fixed by adjusting fan curves or making sure AIOs are changing fan speed off the coolant temp instead of CPU temp. AMD changing the boost behavior to mollify people freaking out over numbers they don't like that aren't actually a problem is lame as hell. Edit: lmao that idle clocks have increased so now they will actually be using more power at idle, but hey the reported temps and voltages will be lower so the people who stare at hwinfo all day instead of using their computer will be happy. Edit 2: Looks like they've changed the boost delay to be more like Windows Balanced or Ryzen Power Saver. Why not just tell people to switch their plan to one of those instead of changing Ryzen Balanced to match? Theris fucked around with this message at 07:23 on Jul 31, 2019 |
|
#
?
Jul 31, 2019 06:50
|
|
|
This really doesn't matter, the downsides and benefits are both drat small. Who cares?
|
|
#
?
Jul 31, 2019 08:31
|
|
|
The only fix I want is for my x470i / 3700x to cold boot with xmp enabled.
|
|
#
?
Jul 31, 2019 09:34
|
|
|
apropos man posted:That Anandtech article says a system that would previously idle at ~2.2GHz now idles at 3GHz. I don't like this. I'd rather see that poo poo clocking low when my PC is doing nothing.
|
|
#
?
Jul 31, 2019 12:36
|
|
|
MaxxBot posted:AMD actually lowered the performance of their chips in response to stupid people on Reddit complaining about a nonexistent issue, I mean it might be a totally insignificant difference but I still don't like it. quote:Note: People have reported performance degradations with the new driver. These degradations have nothing to do with the new power plan / idle behaviour; it's more likely that this is caused by the temporary workaround for the RDRAND (Destiny 2) issues. One can maintain the original performance with the new Balanced power plan idle behaviour by simply selectively installing the new 5.0.0.0 power plan on top of the 07.07 chipset driver release.
|
|
#
?
Jul 31, 2019 14:39
|
|
|
AMD made like three changes at once there. Turning off the RDRAND instruction appears to be what's causing the performance regression (per Andrei at Anandtech). The chip reporting average package temp instead of hottest core is probably the real fix here, that’s what’s gonna keep your fans from revving at the slightest 1% load. Having a fast speedstep is fine but Robert is deliberately conflating that with problems in idling in general. First off, programs don’t “request” clock-up at all, it’s up to the OS and the processor to determine frequency. The only thing you can do is sleep, but once you exit sleep you have no control over what frequency you’re running at. Blaming this on applications “requesting too much performance” is wildly deceptive. But that’s pretty typical for Robert, his role w/r/t the community is basically damage control/crowd wrangling. Second, he’s framing this as “Zen2 is just so good at speedstepping that it clocks up at the drop of a hat”, but if it’s getting stuck in higher performance states (higher idle power) after AMD reduced speedstep intervals, rather than getting stuck in the lower power stage like people wanted, that is suggestive that what’s really going on is that Zen2 just can’t clock down very far unless it’s actually truly idle with a real 0% load and can actually turn the cores off. You don’t need to clock up to max every times thread is running... or at least no other processor does. That’s always been the problem here, there is no “30% clock mode” on Zen2 like there is on previous gens. By all means roll back the changes to speedstep interval - they’re not an actual fix for Zen2’s broken idle anyway. I’d go as far as to say AMD knows it’s a bad fix (the right fix for most of the problems is to change to average temp rather than hottest core) and it’s a red herring to get people to accept the old behavior as proper, rather than further inquiring why idle still isn’t working right. Maxxbot’s reaction is exactly what they’re trying to provoke. Steam running in the background shouldn’t trigger max boost-up, period. There should be intermediate power states between “completely off” and boosted. AMD is very studiously dodging the question of why it’s boosting straight to max, and instead preferring to talk about how great their speedstep frequency is. Which is great, faster speedstep is good, but why is it stepping between off and max boost on minimal loads, instead of stepping to 30% frequency? Paul MaudDib fucked around with this message at 18:44 on Jul 31, 2019 |
|
#
?
Jul 31, 2019 14:41
|
|
|
I suppose that one good thing about this is that AMD are being active in tuning Zen2 via regular updates? They may have made a questionable move this time but in 3 or 4 weeks the next update may do something with more finesse?
|
|
#
?
Jul 31, 2019 15:50
|
|
|
apropos man posted:They may have made a questionable move this time but in 3 or 4 weeks the next update may do something with more finesse? Squeaky wheel always gets the oil, not much you can do about that, and the stupidity seems easy to avoid if you want for now so its not a big deal. Surprised no one bothered to mention it but 1usmus finally released his Ryzen Memory tuner app for 3xxx series chips. Took longer than expected to come out but it seems to be pretty good so far. Here is the latest version of it. Combined with BZ's information about the ideal targeted clocks and timings from that page up the ways a bit and anyone interested it trying to squeeze the last few percent of performance out of the RAM should be set if they want.
|
|
#
?
Jul 31, 2019 18:34
|
|
|
I guess Lisa Su mentioned yesterday that Ryzen 3000 is selling 3.5x faster than Ryzen 2000.
|
|
#
?
Jul 31, 2019 22:30
|
|
|
3.5... not great, not terrible. Investors: “the readout only goes up to 3.5” Paul MaudDib fucked around with this message at 01:26 on Aug 1, 2019 |
|
#
?
Jul 31, 2019 22:34
|
|
|
Great numbers, now they just need to pull ahead in graphics cards as well.
|
|
#
?
Jul 31, 2019 22:36
|
|
|
Lambert posted:Great numbers, now they just need to pull ahead in graphics cards as well. I'm not holding my breath on windows drivers fixes. AMD tends to only fix game related issues and lets the windows desktop issues rot away. The big problems with the card's drivers are present with their last gen cards too. Khorne fucked around with this message at 23:40 on Jul 31, 2019 |
|
#
?
Jul 31, 2019 23:35
|
|
|
Paul MaudDib posted:3.5... not great, not terrible.
|
|
#
?
Jul 31, 2019 23:45
|
|
|
Khorne posted:If they fix windows drivers and the AIB cards are decent the 5700XT is one of the most appealing price:perf GPUs in the current market. It's $150 less than 2070 super at microcenter and goes head to head in a decent number of titles. The AIB cards can squeeze decent extra performance as well. I asked at Microcenter about 5700XT AIBs and they expect them to be $50-100 over msrp, at least the ones they're going to carry. Not that microcenter is the best place to buy a gpu.
|
|
#
?
Jul 31, 2019 23:50
|
|
|
MaxxBot posted:I guess Lisa Su mentioned yesterday that Ryzen 3000 is selling 3.5x faster than Ryzen 2000. jfc, all those 3700x sales and fewer 3600 ones than 3900x. *thinks a bit* jfc, I was set to buy a 3700x before seeing reviews.
|
|
#
?
Aug 1, 2019 10:55
|
|
|
That's revenue instead of numbers. That probably still means maybe double the 3700X sales vs 3600, but fewer 3900X sales. It's still a lot more than I expected, but I guess it's early adopters and later sales will probably skew lower.
Stickman fucked around with this message at 11:07 on Aug 1, 2019 |
|
#
?
Aug 1, 2019 11:04
|
|
|
Still a work in progress, lots of cable management to do, need to peel the plastic off the top and front glass panels, and a new gfx card next month, but here's my new 3700X rig  There is a rocket Orc in the case because of the Mantic thread. Compared to the past  I am pleased with my upgrade, carry on AMD
|
|
#
?
Aug 1, 2019 13:38
|
|
|
Stickman posted:That's revenue instead of numbers. That probably still means maybe double the 3700X sales vs 3600, but fewer 3900X sales. It's still a lot more than I expected, but I guess it's early adopters and later sales will probably skew lower. To be fair, they could have sold twice, three times as many 3900x if there was sufficient supply (in Europe). According to their product listing, Mindfactory has sold 1,370 3900x and it has been out of stock for... 3 weeks now? I talked to a sales rep, asked her if the shipping they are expecting on 8/14 covers all the pending up orders and they couldn't tell, they had >1,000 back orders and ordered twice that many (but are not sure if they'll get them all).
|
|
#
?
Aug 1, 2019 13:49
|
|
|
So on the Linux side of things...CentOS 7/7.5 a horrible idea on a 3900X? Looks like there's that systemd and/or RdRand thing, but as far as I can tell BIOS / uCode updates have triaged that problem for now. More concerned about kernel 3.10 and a brand new uArch, and don't want to really recompile and run my own. Also kind of considering actual RHEL 8 since CentOS 8 isn't out yet and this is a work expense.
|
|
#
?
Aug 1, 2019 18:58
|
|
|
movax posted:Also kind of considering actual RHEL 8 since CentOS 8 isn't out yet and this is a work expense. While you’re not supposed to use it as an actual enterprise profit-generating machine, there’s a free “developers license” to RHEL that you can use to make sure there’s not any wacky Ryzen 3000 related bugs before spending $180. When CentOS 8 eventually releases, most /home related stuff should transfer to it perfectly.
|
|
#
?
Aug 1, 2019 19:25
|
|
|
movax posted:So on the Linux side of things...CentOS 7/7.5 a horrible idea on a 3900X? Looks like there's that systemd and/or RdRand thing, but as far as I can tell BIOS / uCode updates have triaged that problem for now. There's some errata fixes for Zen2 in the latest kernel patch that came out this week, that should hopefully address the most egregious stuff for the moment. But I would go RHEL 8 if I had the choice.
|
|
#
?
Aug 1, 2019 19:27
|
|
|
BangersInMyKnickers posted:There's some errata fixes for Zen2 in the latest kernel patch that came out this week, that should hopefully address the most egregious stuff for the moment. But I would go RHEL 8 if I had the choice. Like mainline mainline kernel, or actually for CentOS 7.x series right now? Never setup RHEL before but would definitely give it a try.
|
|
#
?
Aug 1, 2019 22:21
|
|
|
3.10 kernel in 7.6 in RHEL, I assume its available on CentOS at this point since its been a few days.
|
|
#
?
Aug 1, 2019 22:24
|
|
|
BangersInMyKnickers posted:3.10 kernel in 7.6 in RHEL, I assume its available on CentOS at this point since its been a few days. Awesome — I tried poking around the CentOS mailing lists, do you have a link to that patch / the mailing list it would be on per chance?
|
|
#
?
Aug 1, 2019 23:13
|
|
|
I also completed my build, went with the 3600 in the end. Had an absolute nightmare getting it all in, I'm well out of practice after not building a PC for like 8 years. Had a bunch of trouble getting the AIO mounted properly, had to take it on and off a couple of times which smudged the paste. Don't think it's caused me too many problems though. I was hitting 85c in Prime95 SmallFFT, but was at 65c stable for 2 hours of Battlefield V, with some random spikes to 70. I have PBO and AutoOC+200 turned on, but it didn't boost past 4Ghz and hovered around 1.4vCore the whole time I was playing with the newest BIOS on an Aorus Elite X570 - don't know if that's expected or if somethings up. I got an ok overclock on my Micron e-die. Went straight for 3600CL14 using the RAM calculator but it was constant memory errors when testing - I realised I have no idea about memory overclocking so I wouldn't even know where to start to try and make it stable, so I settled on 3466CL14. My fps was significantly better in BFV though, i'm capped at 138 and pretty much never dropped below 100. I'm 90% certain they were all GPU limited situations, because looking at the sky or a wall caused it go back up - at least I hope it is anyway. Don't know if i'll regret not going for the 9700k but even that would've been £250-300 more expensive. Probably going to replace the 2 stock fans that came with the O11 Air because their min RPM is 900 and they have a horrible noise profile. Think I'll to have to get a Commander Pro as well because the Gigabyte fan control software is garbage and can't even have a fan curve run off a GPU temperature reading. Bit of a waste for only 2 or 3 fans though, the 3 that come with the H150i all plug into the pump and I don't think you're meant to put them in the Commander Pro. I've got them at 0 RPM unless a PCI-E sensor on the motherboard hits like ~45 degrees but that temperature doesn't correlate with the actual GPU temps at all so it's pretty poo poo.
|
|
#
?
Aug 1, 2019 23:31
|
|
|
movax posted:Awesome — I tried poking around the CentOS mailing lists, do you have a link to that patch / the mailing list it would be on per chance? Kernel 3.10.0-957.27.2 is the one with some zen2 fixes in it bundled in with CVE stuff. It's all errata fixes so don't expect any documentation for it. Have a support case open with right now dealing with some zen stuff which is how I found out
|
|
#
?
Aug 1, 2019 23:39
|
|
|
Stickman posted:That's revenue instead of numbers. That probably still means maybe double the 3700X sales vs 3600, but fewer 3900X sales. It's still a lot more than I expected, but I guess it's early adopters and later sales will probably skew lower. It's also individual CPU sales from one German enthusiast retailer if I understand correctly. That data is probably indicative of what self-building pc enthusiasts are buying in general, but that's a tiny piece of the market. I'd be real wary about extrapolating anything about company performance and especially anything about the actual sku breakdown.
|
|
#
?
Aug 2, 2019 00:42
|
|
|
BangersInMyKnickers posted:Kernel 3.10.0-957.27.2 is the one with some zen2 fixes in it bundled in with CVE stuff. It's all errata fixes so don't expect any documentation for it. Have a support case open with right now dealing with some zen stuff which is how I found out Ahh, got it — thanks for the inside scoop! Excited to see how fast Buildroot turns around Linux builds and Vivado builds FPGA bitstreams with 24 threads.
|
|
#
?
Aug 2, 2019 00:45
|
|
|
movax: While you’re not supposed to use it for your business work, using the free “developers license” of RHEL 8 to investigate compatibility with your new hardware isn’t outside the terms of that license. You can just move up to a paid license if you like what you see.
|
|
#
?
Aug 2, 2019 07:03
|
|
|
Bad news for AMD on the laptop front. I knew they should have gotten new APUs out sooner.... https://www.notebookcheck.net/Our-first-Ice-Lake-Core-i7-1065G7-benchmarks-are-in-and-they-outgun-the-AMD-Ryzen-7-3700U.428323.0.html
|
|
#
?
Aug 2, 2019 07:24
|
|
|
SwissArmyDruid posted:Bad news for AMD on the laptop front. I knew they should have gotten new APUs out sooner.... OEMs aren't going to put lpddr4x into mainstream laptops
|
|
#
?
Aug 2, 2019 08:45
|
|
|
Yeah that is the highest end Iris Pro model I think (bigly $$$) stuck with insane botique memory. And it still gets its rear end beat by the mobile 1030/MX150 thing. No idea who this (or APUS) is for.
|
|
#
?
Aug 2, 2019 09:05
|
|
|
SwissArmyDruid posted:Bad news for AMD on the laptop front. I knew they should have gotten new APUs out sooner.... If they were cheap or at least more affordable I'd agree but they're actually fairly pricey + require pricey RAM to perform well and overall don't usually perform significantly better than existing Whiskeylake parts despite the impressive IPC gains because Intel's 10nm still sucks and base clocks on nearly all the parts are at or below 1.3Ghz. Boost clocks on paper are nice of course but on the thermally limited platforms these will all be put in they're not really interesting or much useful in real world use I bet. So I don't think Icelake is going to do much of anything to improve the situation for Intel on laptops. AMD's Ryzen3xxx APU's (still Zen+ core) seem to be mostly targeted at the lower and maybe mid laptop range and do seem to be getting some traction there as a value buy. Zen2 based APU's are supposed to be coming in Q4 2019 so whatever advantage you might argue Intel has here is probably going to be very short lived anyways. sauer kraut posted:No idea who this (or APUS) is for. These new Icelake parts are supposed to do the same thing but the prices are too high IMO to really be practical for someone who wants something like that. The cheapest laptops these seem to be going into, at least around launch until Q1 2020, seem to be priced around $1,000 to start. So if you wanted one with the CPU used in that notebookcheck review you're probably going to have to spend around ~$1,300-1,400 since its the near top end Icelake model they're showcasing there. Other Icelake laptop models seem to be targeting $1,500 to start mind you and will probably go for close to $2,000 with the same CPU model. You can get laptops with the 3700U in them now that start at ~$700-800. And the 3700U is the top end and most expensive option for AMD on laptops right now. If you wanted something more affordable (and reasonable if you're on a tight budget) you'd go for the cheapest 3500U laptops which sell for ~$500-600 typically. To be fair the laptops chassis these Icelake chips are going into are actually pretty nice and probably have at least somewhat better build quality than the cheaper 3700U/3500U based laptops but for many price is going to be one of the biggest determining factors and price-wise there isn't a valid comparison between the cheapest laptops that have either CPU. They're really targeting 2 different markets. Icelake is clearly going for the higher end market and the 3xxx APU's for the low to mid range. The APU's pricing makes sense given their target market and performance but Icelake's really doesn't. PC LOAD LETTER fucked around with this message at 10:10 on Aug 2, 2019 |
|
#
?
Aug 2, 2019 09:34
|
|
|
|
| # ? Apr 18, 2024 02:48 |
|
|
Arzachel posted:OEMs aren't going to put lpddr4x into mainstream laptops They still put LPDDR3 into new notebooks with gen 8/9 Intel CPUs so yeah. It's nice on paper but it's not going to be in many (actually portable) notebooks, which is where a good iGPU or APU makes the most sense (in everything bigger you can just stick an actual mobile GPU and that will be miles ahead).
|
|
#
?
Aug 2, 2019 10:19
|
|
