|
Sweet thanks guys! Didn't know if going Intel -> AMD was gonna futz with it. Worse comes to worse I can wipe it and start from scratch later but would be nice to save some of the time initially. I do have the latest BIOS on a USB drive for my ASUS ROG STRIX z570-e as I can flash it from that button on the back. That should be good to recognize the 5900x yeah?
|
#
?
Jul 30, 2021 21:52
|
|
|
|
| # ? Jun 5, 2024 19:28 |
|
|
I did wipe my installation when I got a new Mainboard. Most annoying part was getting Microsoft to accept my windows key since they tied it to your core components. So that annoyance might be a reason to first install the CPU and see what happens.
|
|
#
?
Jul 31, 2021 00:34
|
|
|
haha yes, our windows keys. legal and true. haha. those we all have as well. haha. laughing
|
|
#
?
Jul 31, 2021 03:58
|
|
|
Just booted and it it's working... *knocks on wood*.  *edit* it's fast
|
|
#
?
Jul 31, 2021 04:06
|
|
|
PC LOAD LETTER posted:There should be at least 1 more major performance boost for AM4 coming.... Really? Nice! I thought Zen3 was the last hurrah. That's good news.
|
|
#
?
Jul 31, 2021 05:54
|
|
|
Cygni posted:haha yes, our windows keys. legal and true. haha. those we all have as well. haha. laughing They tied that to your email years ago didn't they My windows key is from a years-old W7 purchase that's activated just by logging into my microsoft account on re-installs then literally never logging in ever again
|
|
#
?
Jul 31, 2021 10:38
|
|
|
Zedsdeadbaby posted:They tied that to your email years ago didn't they haha yes, logging into a microsoft account, yes of course but yeah im one of those freaks who tries desperately not to give microsoft (or google) any money or information if i can avoid it Cygni fucked around with this message at 10:44 on Jul 31, 2021 |
|
#
?
Jul 31, 2021 10:41
|
|
|
$5 Enterprise keys off Ebay so you can set group policies the  Professional version has been ignoring since 2015 or whatever.
|
|
#
?
Jul 31, 2021 10:56
|
|
|
orcane posted:$5 Enterprise keys off Ebay so you can set group policies the I have been doing this. Cheap and you get the most comprehensive version of win10.
|
|
#
?
Jul 31, 2021 13:16
|
|
|
Cygni posted:haha yes, logging into a microsoft account, yes of course I completely understand and empathize with this, it is why I have no social media at all, and immediately jailbreak my phones My microsoft account is not the same as my main email for the same reason, it's just to access microsoft services and absolutely nothing else.
|
|
#
?
Jul 31, 2021 15:52
|
|
|
You can still weasel out of signing up for an account if you unplug Ethernet/Wifi, right?
|
|
#
?
Jul 31, 2021 15:56
|
|
|
Word of warning, don't intend to do a fresh install upgrade to Windows 11, because that trick don't work anymore apparently.
|
|
#
?
Jul 31, 2021 16:07
|
|
|
Nomyth posted:You can still weasel out of signing up for an account if you unplug Ethernet/Wifi, right? Right now yes, but it's been announced Windows 11 home edition will require internet access & an ms account to install. Pro or better should still let you avoid it.
|
|
#
?
Jul 31, 2021 21:01
|
|
|
Has anyone had trouble with getting an AMD RAID array to show up in Linux? I have an Aorus Pro X570 with a single RAID 10 array set up, however Centos 8 is not picking up the array - it only sees the disks directly. I had it running fine on F33h, however after upgrading to F34 the array was invalidated or corrupted (unfortunate, but all the data was backed up anyway), so I removed and recreated it. However, the new array is not picked up by mdadm even after scanning for it. Not many results online for this situation - most seem to recommend using rcraid instead, but the drivers appear to be significantly out of date.
|
|
#
?
Aug 1, 2021 09:51
|
|
|
Any particular reason you're using the UEFI-configured RAID rather than plain Linux md?
|
|
#
?
Aug 1, 2021 14:44
|
|
|
SamDabbers posted:Any particular reason you're using the UEFI-configured RAID rather than plain Linux md? No strong reason - I thought that AMD RAID may offer an advantage over software RAID, but if that's not the case then I'll go with mdadm.
|
|
#
?
Aug 1, 2021 15:24
|
|
|
In my experience, RAIDs done in the BIOS have the same amount of resource utilization as pure software methods, but have the immediate downside of often disabling/hiding the SMART status of the attached drives so the OS cannot sound the alarms when one of them is degrading or about to fail. (Not that SMART always catches when a drive is on its way out, the sudden catastrophic failures usually evade it, but on the other hand it will alert you for basically every graceful failure mode and often signal it days/weeks before you would begin to lose data allowing ample time to backup and replace the drive.)
|
|
#
?
Aug 1, 2021 15:33
|
|
|
Bjork Bjowlob posted:No strong reason - I thought that AMD RAID may offer an advantage over software RAID, but if that's not the case then I'll go with mdadm. Tl;dr: You're better off using plain mdadm. The UEFI chipset RAID is still a software driven RAID that requires the OS to have a driver that supports the UEFI's metadata flavor. The big downside of this compared to plain mdadm with standard Linux metadata is that it's more difficult to troubleshoot and usually not portable between UEFI implementations. I recommend turning off RAID mode in the UEFI and just going with plain mdadm.
|
|
#
?
Aug 1, 2021 15:51
|
|
|
Nomyth posted:You can still weasel out of signing up for an account if you unplug Ethernet/Wifi, right? as of 21h1 yes, I just installed a (vm) yesterday with the network disabled and it whined and cajoled me to submit my personal information and invite cortana into my life and also submit my biometrics because of how much money they can make off them, but in the end it still had a "i'm literally in the middle of nowhere with no internet connection just let me loving install" button. How the hell are enterprise keys so cheap? Even in tertiary markets corporate licenses should go for more than a handful of whatever un-valued currency they use.
|
|
#
?
Aug 2, 2021 10:19
|
|
|
Harik posted:How the hell are enterprise keys so cheap? Even in tertiary markets corporate licenses should go for more than a handful of whatever un-valued currency they use. Microsoft doesn't care about making money off the end user, and in an enterprise setting, they'll just audit the company and ding them tens thousand of dollars per VM for Windows Server licensing non-compliance.
|
|
#
?
Aug 2, 2021 10:28
|
|
|
Kazinsal posted:Microsoft doesn't care about making money off the end user, and in an enterprise setting, they'll just audit the company and ding them tens thousand of dollars per VM for Windows Server licensing non-compliance. which would explain home and pro, even ultimate (do they make ultimate anymore?) but not enterprise keys. If I can't buy them and use them for my business without being out of compliance then they're not legitimate keys.
|
|
#
?
Aug 2, 2021 12:43
|
|
|
The same idea applies though, Microsoft largely doesn't care about end users messing around with the enterprise edition, while real enterprises will have to pay up if they wanna use it anyway. A handful of users and mom & pop shops using grey/black market Windows 10 Enterprise keys probably doesn't even register on their radar. So the keys are only legitimate in the sense that Windows will activate, but people who use them obviously have no guarantee it stays that way and they can't be sure they won't lose Windows activation at some point (eg. if the volume license the key was part of gets blacklisted) but at literally $5 a key that's not a big problem. orcane fucked around with this message at 16:55 on Aug 2, 2021 |
|
#
?
Aug 2, 2021 16:45
|
|
|
I know Intel has some good bits and bobs on they're iGPUs for Plex and media transcode, how are Vega iGPUs on Ryzen embedded/APUs? Just mulling my NAS upgrade.
|
|
#
?
Aug 2, 2021 17:55
|
|
|
NewFatMike posted:I know Intel has some good bits and bobs on they're iGPUs for Plex and media transcode, how are Vega iGPUs on Ryzen embedded/APUs? Just mulling my NAS upgrade. Plex itself does not support AMD hardware encoding on their Linux build - the AMD linux driver supports it and the Windows build of Plex supports it, but the Plex build for Linux doesn't support it. That's the biggest problem, as most of the people running plex servers are doing it on Linux or linux/unix based appliances. Other than that it's fine, but it's the same old Vega, transcode quality is not all that great. Their H264 encoder sucks even on current-gen products, and RDNA is where AMD updated the H265 encoder while Vega still has the older crappy one. But then, QuickSync isn't all that great either, of course (but still better than AMD AMF). NVIDIA was solidly ahead even before Turing, and Turing's NVENC dumps on everything else on the market. It would have been really nice if their Zen3 APUs at least supported RDNA1. The RX 5000 series GPUs came out like a year before Zen3 so I don't really get the "RDNA1 wasn't ready / the timing didn't work out" argument some people make. Paul MaudDib fucked around with this message at 20:21 on Aug 2, 2021 |
|
#
?
Aug 2, 2021 19:24
|
|
|
That's a bit of a bummer, but does inspire the thought of running a NAS off an Nvidia SBC that they market for edge stuff 
|
|
#
?
Aug 2, 2021 20:07
|
|
|
NewFatMike posted:That's a bit of a bummer, but does inspire the thought of running a NAS off an Nvidia SBC that they market for edge stuff out of curiosity, are you referring to something specific here? there's the Jetson boards but unfortunately tegra tends to run behind on architecture relative to desktop - right now the most advanced Jetson is the Xavier with a Volta generation gpu aboard, and Ampere based SOCs are sampling this year for launch next year. I still kick myself for not picking up one of those surplus rabb.it servers with like 8 jetson boards and a NUC in a 1U chassis for like $200. That would have been an interesting toy. They were only Jetson TK1s (Kepler) though.
|
|
#
?
Aug 3, 2021 03:26
|
|
|
Yeah, I was looking at the Jetson boards, I didn't realize they were so far out of date. Although my Celeron Synology NAS from like five years ago isn't the speediest thing in the world.
|
|
#
?
Aug 3, 2021 04:54
|
|
|
New APUs codenamed "Cyan Skillfish" were spotted in the Linux driver, which are Navi 1x parts. (DING DONG, loving VEGA IS DEAD.) That's of little consolation to you now, but we shouldn't have to deal with GCN for much longer, thank god.
|
|
#
?
Aug 3, 2021 05:34
|
|
|
NewFatMike posted:Yeah, I was looking at the Jetson boards, I didn't realize they were so far out of date. Although my Celeron Synology NAS from like five years ago isn't the speediest thing in the world. plex is a binary distribution (because it's proprietary) right? getting it to work on Jetson also might be a problem, both the application itself and transcoding specifically (I read some forum comments that suggest the standard NVENC api doesn't work, not sure if that's true/if it still is a problem). I heard Linux 4 Tegra is good but I've never tried it - apparently it is "basically Ubuntu 14.04 with pre-configured drivers for bootloader, kernel, OpenGL, X.Org, Multimedia, etc", maybe you could coax the Plex binary for Ubuntu 16.04 Armv8 to run on it. one of the open-source ones like jellyfin might be easier there, but still could potentially have problems with transcoding (if the API is different). this is one of the reasons I think Gemini Lake is really not a bad little solution - it runs standard x86 binaries/normal distros, and intel does a really great job with their *nix open-source drivers (and you can even pass through the iGPU into a VM container and run the video encoder inside a guest...). I see some new prebuilt NASs out there with like J4115 and J4125 and poo poo and those are really cute little boxes for an all-in-one NAS+plex server. Jellyfin appears to have support for the Raspberry Pi 4 and its hardware transcoding though. No idea where that falls into the NVIDIA-Intel-AMD hierarchy for video quality, but if you're not attached to Plex that would be an option. Paul MaudDib fucked around with this message at 06:10 on Aug 3, 2021 |
|
#
?
Aug 3, 2021 06:07
|
|
|
setting aside the quality of AMD's encoder... * do all AMD GPUs have it? I know that NVidia gates NVENC behind the GTX SKU, such that the GT 710 / 730 / 1030 wouldn't have it, and you have to go with at least a GTX 1050 / GTX 1650 minimum for a Pascal NVenc, or at least a GTX 1650 Super for a Turing NVenc, but what about AMD? Would an RX 550 still have their encoder? * do all AMD APUs have it? would an Athlon 3000G still have an encoder?
|
|
#
?
Aug 3, 2021 06:30
|
|
|
Nobody has mentioned it yet, but the 5600G and 5700G left embargo today: https://www.youtube.com/watch?v=8cH8gwcVqPY https://videocardz.com/newz/amd-ryzen-5000g-cezanne-apu-review-roundup if you HAVE to game on an IGP for some reason, its the best one out there, but its kind of in a weird price position vs both the other Zen 3 parts and Intels 10th/11th gen.
|
|
#
?
Aug 3, 2021 22:40
|
|
|
Cygni posted:Nobody has mentioned it yet, but the 5600G and 5700G left embargo today: haven't watched the video but working from the Tom's Hardware review that went live a few weeks ago: the market competitor for the 5700G ($359) is the 11700 ($339) or 10700F ($259) and the competitor for the 5600G ($259) is the 11400 ($190) or 11400F ($185) or 10400 ($165) or 10400F ($151). it goes without saying that Skylake is still a good deal if you don't need a powerful iGPU, or especially the F models with no iGPU at all. if you need a powerful iGPU, on the intel side you pretty much have to go 11th gen. And you will always be stuck with the Intel drivers - they do basically have to improve them if they want to be taken seriously in the dGPU market, and all the Xe GPUs are architecturally similar, so I think the situation will improve over time, but they also probably will never get the depth of fixes on older titles that AMD has had the chance to build. AMD is more power efficient - the Intel results fall somewhere between "slower when TDP limited" and "burns an insane amount of power just to match AMD when TDP limits are removed". And I think that is actually a big problem for a lot of the target markets for these processors - if you were building a SFF PC then the difference between a 160-200W processor and a 90W processor (actual boost power draw/PPT, not the fake TDP amd advertises) is actually a substantial one in terms of cooling/noise. And while I recognize that the beige-box/OEM-prebuilt market has more concerns than just the technical data - pulling twice as much power under load, or being slower at the same TDP, is not exactly a selling point for OEMs either. All things (price) being equal, over time the market will eventually shift to the more efficient offering. I'm not impressed with AVX-512 on Rocket Lake, it doesn't seem to have produced noteworthy performance gains even in the ideal scenarios - AMD literally pulls half the task energy of Intel in x265 encoding, for the same encode times (meaning - Intel's power draw is twice AMD's for no benefit - and if you power limit Intel they'll fall way behind), and that's basically the best case scenario for AVX-512 gains. It'll probably do better on the nodes for which it was designed from scratch, rather than the backported mess that Rocket Lake seems to be, but it's not really a selling point in this situation, and in fact on Rocket Lake it's just not... good. imo considering all these factors, the 5700G is priced OK for what it is - compared to the 11700, you get a better iGPU, AMD drivers, and better power efficiency, for roughly the same price (I don't think it's fair to quibble over 20 bucks on a $350 processor, they are basically the same price segment). However both the 5700G and 11700 are embarrassed by the 11400, the 5700G ends up being almost twice the price for 2 additional cores and in a vacuum I think that is a tough sell. And the 5600G is not price competitive with the 11400, it needs to come down substantially. Which is a shame because if it were price competitive that would be the standout of the AMD lineup I think - it maintains almost all the clocks of the 5700G, the same cache size (minus 2 cores' worth of L2, which isn't accessible across cores anyway, but L3 is the same), and only loses 1 CU from the graphics. I'm guessing the higher cache-per-core and higher TDP-per-core will mostly cancel out the slightly lower peak clocks so it basically winds up performing the same as 5700G per-core but only with 6 cores. the 5300G is clearly targeted downmarket, right now that segment is very very neglected and has to buy overpriced 3000G and 3200G/3400G crap (3000G was $130 last time I checked - absurd and terrible). It's got much less cache and only 4C8T, which is fine for the sub-$150 segment but not really an enthusiast part that directly substitutes for the 5600G/5700G. imo if you intend to transition to a dGPU down the road, it seems like Intel is the way to go there. And yes, you should give serious consideration to buying a 3600/3700X/5600X/5800X or 11400F instead, with a super cheap dGPU, if that's your scenario, because if you feasibly can run a dGPU then even a poo poo-tier RX550 or something will pants the APUs, it's worth thinking about whether you could make that work instead of gimping your dGPU performance permanently because of the lower cache/IPC on these APUs. And if you are willing to go with an iGPU-less system, the Intel -F chips (both 10th and 11th gen) are very good value while performing equally well in gaming for the most part. The 5700G is 40% more expensive than the 10700F (gaming dGPU build), twice as expensive as the 10400 (office build), and more than twice as expensive as the 10400F (budget gaming dGPU build), and the 5600G does not compare favorably either in terms of pricing. For value, the 11400/11400F and 10400/10400F are probably the best deal on the market in hellworld, by a pretty substantial margin. idk, I'm still very interested in the 5700G for a SFF mini desktop build (iGPU only in a super tiny case like Mini Box M350 or Deskmini X300 with a Noctua L9a) but I'm primarily interested in the iGPU, the power efficiency, and the drivers. And if the 5600G were $200-210 instead of $260 I'd probably go with that instead. And I wonder if Zen3+ will include chips with cache dies stacked on these APUs (not just the enthusiast chips), as that would substantially improve IPC and iGPU performance - but maybe the APUs weren't designed to support that, not sure. Maybe I should just wait and see what happens with Zen3+, it probably won't be that far off given that it's supposed to be a Q4 launch and we're midway through Q3. Maybe another October announcement and November availability. But then, there's no guarantee stacked-cache APUs launch with the rest of the chips anyway - they could be launched separately (if they exist at all). Paul MaudDib fucked around with this message at 06:26 on Aug 4, 2021 |
|
#
?
Aug 4, 2021 00:40
|
|
|
Paul MaudDib posted:haven't watched the video but working from the Tom's Hardware review that went live a few weeks ago: I think the Igpu does provide some value in gaming in that you can use that as your obs encoder, so your cpu and gpu don't get taxed, and if you tell chrome to be on "power saving" via windows and use h264ify chrome extension then the igpu also gets used for youtube watching.
|
|
#
?
Aug 4, 2021 01:21
|
|
|
gradenko_2000 posted:setting aside the quality of AMD's encoder... see the "video encoder" field under "APUs" and "GPUs". It would appear anything GCN and newer has a video encoder, along with TeraScale 3 APUs (but not dGPUs). I actually tried to just swap in one of my spare cards and give it a try and unexpectedly had a whole ordeal. I can't get the AMD drivers under windows to recognize my RX460, even if I go back to an older version where it wasn't a "legacy/unsupported" card. It'll install but I can't launch the control center. Tried DDU, tried reinstalling windows, no luck. I wonder if maybe this RX 460 was used for mining and it has a flashed VBIOS and that's why it isn't recognizing it? You would think the drivers would recognize it the same either way... Weird. But it looks like it also isn't recognizing some of the X99 chipset drivers so it might be a borked installation that I just need to fix in general. But I had no problem running my 3090 on that system for a while till my main gaming rig was back up and running. Works OK under Ubuntu 20.04 but I couldn't get Handbrake to see it (possibly not built with support), but I could manually invoke ffmpeg with h264_vaapi and it seems to be working (although I haven't checked the output yet) Paul MaudDib fucked around with this message at 03:49 on Aug 4, 2021 |
|
#
?
Aug 4, 2021 03:41
|
|
|
wargames posted:I think the Igpu does provide some value in gaming in that you can use that as your obs encoder, so your cpu and gpu don't get taxed, and if you tell chrome to be on "power saving" via windows and use h264ify chrome extension then the igpu also gets used for youtube watching. sending back a compressed stream (NVFBC+NVENC) rather than an uncompressed stream (OBS Studio) is preferable though, because video captures does congest the PCIe bus as well. 50 mbit/s for a compressed stream, not so much, but an uncompressed bitmap stream is (eg) 3840x2160 x 24 (8bpc) = 200 mbit per frame, so at 60fps that's 12gbps or 1.5 Gbyte/s. Your APU only has PCIe 3.0x16 or maybe 3.0x8, which is 16 or 8 Gbyte/s (easy trick: PCIe 3.0 has 1 GB/s per lane). It's not gonna singlehandedly wreck frametimes, but it's also a non-trivial chunk of the pcie bandwidth. video playback makes sense I guess, as does office PC use that just needs "anything", and they're nice to have for debugging/etc. Not saying there's no uses, just that iGPUs in a PC with a dGPU aren't necessarily the answer, especially when (as with the AMD APUs) they come with a downside relative to the "enthusiast" versions. Paul MaudDib fucked around with this message at 06:37 on Aug 4, 2021 |
|
#
?
Aug 4, 2021 06:24
|
|
|
I watched HWUB's review of the 5600G and was unimpressed. 260 USD for something that can't beat an i5-10400 is not great. I'll keep holding out for an even cheaper gateway to Zen 3.Paul MaudDib posted:see the "video encoder" field under "APUs" and "GPUs". It would appear anything GCN and newer has a video encoder, along with TeraScale 3 APUs (but not dGPUs). Thanks for checking this out, I appreciate it.
|
|
#
?
Aug 4, 2021 06:52
|
|
|
Hello I am dumb. I just purchased a Ryzen 5 2600X and I have crucial DDR4-3600 RAM. But reading up on things, it seems like this ram is too fast? Like I shouldn't go above 2933? Does that mean using this ram is dangerous? Should I just buy smaller ram, or can I install this and it will simply run slower?
|
|
#
?
Aug 4, 2021 06:58
|
|
|
Skyarb posted:Hello I am dumb. I just purchased a Ryzen 5 2600X and I have crucial DDR4-3600 RAM. But reading up on things, it seems like this ram is too fast? Like I shouldn't go above 2933? Does that mean using this ram is dangerous? Should I just buy smaller ram, or can I install this and it will simply run slower? there is no such thing as "too fast" RAM in terms of physically breaking your computer your motherboard/CPU will always find a way to run/use your memory, even if it's slower than the rated speed, without you having to futz with it you don't need to buy different RAM
|
|
#
?
Aug 4, 2021 07:02
|
|
|
Skyarb posted:Hello I am dumb. I just purchased a Ryzen 5 2600X and I have crucial DDR4-3600 RAM. But reading up on things, it seems like this ram is too fast? Like I shouldn't go above 2933? Does that mean using this ram is dangerous? Should I just buy smaller ram, or can I install this and it will simply run slower? Everything above 2400 is out-of-spec for DDR4. The only question is how high an overclock you can push. Like gradenko_2000 said, your RAM will be fine. Run it at whatever your MB will accept, and if you have problems with stability back it off to 3200 or so. I had to run 3000 on zen+, but that's with 48gb made of 6 sticks of ram, a bit unbalanced.
|
|
#
?
Aug 4, 2021 15:30
|
|
|
|
| # ? Jun 5, 2024 19:28 |
|
|
Harik posted:Everything above 2400 is out-of-spec for DDR4. 
|
|
#
?
Aug 4, 2021 15:49
|
|