Ranter
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Comments
-
kamen69252yThat's a little too generalised.
I'm on the AMD side for the CPU and there were some micro-stuttering previously, but that was fixed with new motherboard firmware (so check for updates for your board). Can't say anything about the GPUs though - I haven't touched an AMD GPU in a while. -
What Ryzen 9 is that? 3900X? There was some bios setting for the idle current because shit would black out during idle phases. IIRC, you'd need to set Power Supply Idle Control to "typical".
-
As for GPUs, AMD has quality problems with their drivers and never got their shit together.
The only point where you can buy an AMD GPU is shortly before the introduction of a new generation. By then, even AMD should have fixed their drivers (or don't buy if they havn't), and you get considerable price reduction.
Given the teething issues they had also with Ryzen (took several bios updates), just don't buy any AMD products right after launch. You pay more money for a bad experience.
After that however, you can get quite good bang for the buck.
B450 / 5700G / 6650 XT here, i.e. a full AMD rig. -
@kamen @kamen No they were are not fixed. Just playing an 8 hours video on youtube, it's a garanty it's gonna stutter from time to time. Like 100%.
System 100% up to date.
Like in chrome (edge). Only like 40-50 tabs opened, switching to another tab takes sometimes a whole 5-7 seconds.
And I have noew 64GB ram (vs 16 before).
Ok, granted, I have a LOT of things lunched, but they were all stabble with intell/nvidia. -
@NoToJavaScript Did you check out that bios setting?
Also, my 6650 XT would hang the whole system once per week with kernel 5.15 LTS, but it looks like that's fixed with kernel 6.0 OEM in Mint/Ubuntu. It's crazy how long that took, I fully expected 5.15 to be mature with RX 6000.
If Nvidia had proper open source drivers, I might have considered a 3060 Ti or something, but that's just something AMD has going for them. Also, the 6650 is quite a bit cheaper. RT/CUDA/ML are not really useful on 3060 class cards anyway.
At least on the CPU side, shit is rock stable. Swapped the initial 3400G for a 5700G and sold the old one on Ebay for a crazy good price. -
@Fast-Nop I'm on windows.
The only thing which works GREAT is video games. 60 fps / 4K, sometimes even some ray tracing : No problems (Except GPU reaching 105 degree C).
That's the thing. I expect system to work fine 'as is" (Without ever going into BIOS).
Going into bios should be if you want "more".
I actually went into bios, because my memory was clocked at... 50% of potential. (Also, never happens in Intel world).
No, I'm 38 YO, I expect the Default system work well AND if you want more from system, do fine tunning. I won't spend time playing with BIOS settings, been there, done that, fuck it.
Any APP, System, Hardware should just work fine (With possibility to get more out of it with tunning ofc). -
@NoToJavaScript Even on Intel systems, the memory will not run at more than JEDEC speed out of the box - you have to enable XMP if you build the system yourself. Same with AMD, so that's normal.
That's because most RAM kits are not able to meet the advertised speed and timings at 1.2V, only at 1.35V which technically violates the JEDEC spec and counts as OC, only that's it's manufacturer validated.
The Idle Power Supply setting does suck and shouldn't be required, but I'd give it a try so that you may end up with a more stable system. -
iiii90852yUsing Ryzen 5 1600 for years now. No complaints whatsoever. What are you talking about?
-
Parzi86632yI've been using AMD CPUs/GPUs for years. Never really had an issue. Currently using a Ryzen 7 5700X CPU and an RX 5500XT GPU, and I can do basically anything just fine. Their drivers aren't the greatest on Windows, but I've never really had massive issues with them, just minor things. On Linux, this setup is more or less optimal, though.
Hell, my Llano AMD APU lasted 11 years before actually needing an upgrade, and that was used in a gaming rig! (note: external GPU was upgraded over time in this build, the APU graphics side was usually only used by the OS.) -
Parzi86632y@Fast-Nop I've noticed Ubuntu and derivatives tend to freeze drivers for months at a time, most AMD driver issues are fixed in a week or so most of the time, but a lot of common distros don't ship updates that often. (This is why I prefer Debian's unstable branch, actually.)
-
@Parzi I would agree to that if we were talking about new hardware. However, a 6650 XT having system freeze issues with 5.15 is ridiculous, given how old RDNA2 was already when 5.15 LTS came out.
Also, Ubuntu usually has the HWE kernels (hardware enablement) for new stuff, and that's OK. What's not OK is that AMD needs that even for not-so-new hardware because these dumb motherfuckers have no quality control and crap out any buggy shit.
The whole "AMD drivers" trope is just as bad as it was 10 years ago. It's why Nvidia can command these crazy prices, because AMD is no serious competition until these morons get their shit together. -
@Parzi Just one example. After upgrading from Mint 20.3 to 21 (i.e. Ubuntu 20.04 to 22.04), the 3D performance of my 6650 XT tanked to a quarter.
AMD had uploaded a firmware to the kernel git where the clock under 3D was not ramped up, and that was right when Ubuntu made their freeze. It took me a Ubuntu bug report to get them to update the linux-firmware package for that GPU because AMD did correct it later.
My point: how on earth did some dumb motherfucker at AMD upload a firmware to the kernel git with that issue?
It's because they have not tested that in any way, and that's because AMD has no quality control processes. There should have been a series of tests performed before uploading anything, and that should have been mandated by some sort of AMD internal release process. -
Parzi86632y@Fast-Nop That's true, someone should probably be checking the firmware and such first. I've not seen an issue last longer than a couple weeks, though, so they're at least paying attention, and not being completely dishonest about open sourcing their drivers like Nvidia.
-
@Parzi The number of fuckups that AMD can afford as underdog with like 10% market share for dGPUs is pretty exactly ZERO. They have to execute on spot if they want to convince anyone to change.
As it stands now, people rather pay the Nvidia premium just to not shell out several hundred bucks for a product that may work after some weeks, or may stop randomly and then works again after some weeks and/or fudging around.
Random black screen issues (RX 5700 XT), monitor suspend problems both in Windows and Linux (RX 6000), claiming energy efficiency while guzzling 80W watching Youtube (RX 7000) - you can always count on AMD to fuck up in some way. Plus their current cooler problems.
If their management continues to fail recognizing and fixing their shit, AMD may soon see themselves in the third place for dGPUs at maybe 5%. After Intel, provided that they learn from their atrocious A380 driver disaster. -
Parzi86632y@Fast-Nop A lot of the same issues crop up with companies like Dell, or HP, or most other portable device companies, really. These companies still have massive market share in their fields. I don't know if that's gonna hold true.
Recently had to deal with Dell support, actually. Their premium packages aren't any better than having no warranty at all. -
@Parzi Dell and HP are infamous for shoddy, proprietary hardware - but with dGPUs, people have choices.
As the Steam survey shows, the most popular one is sticking with older Nvidia cards, followed by buying new Nvidia cards in long cycles as reaction to the pricing.
As it stands, AMD fails to convince the first group to instead opt for buying an AMD card, and there are good reasons that AMD continously fails to address. It's as if their management lived in a parallel universe where they ignore that and then wonder why nearly nobody buys their dGPUs.
The current RX 7000 cooler issues are just so typical. Why is it that no sample tests have been conducted even on their so-called top of the line models? Again, because sloppiness is a big part of the AMD corporate culture, and management needs to make changing that a top priority. -
Plus that AMD's pricing is really off. A 7900 XTX at $1000 vs a 4080 at $1200 is dumb because if people are able and willing to shell out that much for a GPU, the extra $200 are no serious consideration.
Plus that the AMD ref model is meh so that you'd have to go for even more expensive AIB partner models while the Nvidia FE is something you would actually buy.
If AMD thinks they can price Nvidia minus some rebate, they're totally off the mark. Half of Nvidia, then we're talking - but AMD somehow thinks they're nearly on a par when in reality, nobody except the AMD management thinks so. -
@Fast-Nop
yes XMP basiclly the only tweak I did manually.
Well, apparently reboot once a month helps, Rebooted this morning, seems to have way less issues.
When you go to "Oh they do it cheap", don't expect results...
Changed my PC build around 2 years ago.
Went from Core i7 / Nvidia to Ryzen 9 / AMD
Welp, AMD is totaly unstable.
I've invested 5k $ so I'm gonna ride it, but NEVER, EVER EVER again I'm buying AMD CPU or GPU.
Shit is unstable as fuck. I have latency issues, CPU issues, Video issues almost every week.
With Intel/nVidia cvombo I had before, I had issues maybe once every 3-4 months.
So yeah, buy low cost AMD, you pay the price later in usability. Fuck them.
rant