AMD Ryzen 9000 Series processors are ready to deliver world class gaming(twitter.com)
twitter.com
AMD Ryzen 9000 Series processors are ready to deliver world class gaming
https://twitter.com/amdryzen/status/1820956835794358451
111 comments
> the speed is DDR5-5600 when 2 modules are installed
That's the conservative JEDEC-compliant speed, but if you want to go for maximum performance they are saying to aim for DDR5-6400 this time, up from DDR5-6000 on the previous generation.
https://www.techpowerup.com/324023/ddr5-6400-confirmed-as-sw...
That speed is likely only attainable with 2 modules installed, but with 96GB (2x48GB) kits available now that's not much of a limitation for most people.
That's the conservative JEDEC-compliant speed, but if you want to go for maximum performance they are saying to aim for DDR5-6400 this time, up from DDR5-6000 on the previous generation.
https://www.techpowerup.com/324023/ddr5-6400-confirmed-as-sw...
That speed is likely only attainable with 2 modules installed, but with 96GB (2x48GB) kits available now that's not much of a limitation for most people.
With enough time sunk into it, these numbers are more... suggestions, anyway.
My system has a 7950X3D and four sticks at DDR5-6600. The module density, board trace topology, and luck-of-the-draw on IMC/SOC can matter a fair bit.
Tuning voltages/timings can make up differences. Also remember the infinity fabric clocks and such being somewhat paired
My system has a 7950X3D and four sticks at DDR5-6600. The module density, board trace topology, and luck-of-the-draw on IMC/SOC can matter a fair bit.
Tuning voltages/timings can make up differences. Also remember the infinity fabric clocks and such being somewhat paired
May I suggest to also run memtest if you're gonna tune these kind of things. Your computer may appear to work normally after changing the settings, but memtest can reveal unnoticeable memory errors.
memtest is a pretty basic test and is not very reliable at triggering errors in memory pushed just past of what it is capable of. It's fine to run it as a smoke test before booting into the operating system (to make sure it won't crash right away), but after that you're gonna need something like TestMem5. I keep a Windows install just for this.
Cool, didn't know about TestMem5 until now. Thanks!
Good suggestion, though I have - Prime95, memtest, many long workloads :)
With DDR5 and really, most modern hardware, having some more robustness (that pseudo-ECC whatever, clock stretching)... it might be worth considering performance as part of your validation
With DDR5 and really, most modern hardware, having some more robustness (that pseudo-ECC whatever, clock stretching)... it might be worth considering performance as part of your validation
Sounds good. I didn't mean to make a suggestion personally to you by the way. Sorry if it seemed that way. English is not my native language. My remark was more targeted towards the reader of the thread who gets curious and may start experimenting.
Totally fair :) I err towards defensive, no harm done
What's the first word latency/CAS latency? I think EXPO 6000CL30 kits can achieve 10ns, so if you have to run 6600CL40 or something it may be worse.
It's CAS 32 - one doesn't have to go sky high latency to get high frequency, the binning is truly incredible these days.
Just a boring Corsair Vengeance kit with OOTB settings as far as they're concerned. The CPU needed help
Just a boring Corsair Vengeance kit with OOTB settings as far as they're concerned. The CPU needed help
Nice, which ICs does the RAM kit have? I don't think your experience is usual by any means with uclk:fclk 1:1 though.
Can't say I've been brave enough to tear them apart :D I took the lazy way out and just sorted by specs on the listing.
I actually expressly didn't want to overclock the memory, just get something that ran nicely as-is.
I'm actually not at 1:1... stable. Perhaps wires got crossed? In another post I mentioned it POSTs but that's about it :( I may be able to get something there, but can't promise or over-represent it.
The memory is at 6600 but I have some divisor or whatever at memclk/2. This is super stable
I actually expressly didn't want to overclock the memory, just get something that ran nicely as-is.
I'm actually not at 1:1... stable. Perhaps wires got crossed? In another post I mentioned it POSTs but that's about it :( I may be able to get something there, but can't promise or over-represent it.
The memory is at 6600 but I have some divisor or whatever at memclk/2. This is super stable
6000 at cl32 is most likely going to be Hynix A die, the cas latency could be better if they push like 1.8V into the memory but it's not a particularly important timing for performance
Is this with the EXPO profiles, or are they running at the JEDEC transfer rate?
This is indeed with EXPO! There were great strides made with firmware revisions in the last year or so.
I do have to make a point to boost the supporting IMC/SOC voltages before enabling that, or I can get a bit wedged.
The profile voltages (and clocks/timings) for the sticks themselves work fine - just the board/CPU itself needs some care.
I do have to make a point to boost the supporting IMC/SOC voltages before enabling that, or I can get a bit wedged.
The profile voltages (and clocks/timings) for the sticks themselves work fine - just the board/CPU itself needs some care.
Is that running at 1:1 uclk/mclk?
Glad you asked! I just checked and tried, no/not yet.
It's at 1:2 or 2:1 or whatever makes sense, my firmware shows memclk/2 - haven't had my caffeine yet.
The memory can train/the system POSTs, but the OS fails to load. With some more fiddling this may be feasible - I'm not sure. Excited to try!
Additionally: the IF frequency is 2100Mhz, trying to go much higher doesn't work out. If I could get 2200 it'd factor nicely into the memory frequency too, but alas - chip isn't up to it.
It's at 1:2 or 2:1 or whatever makes sense, my firmware shows memclk/2 - haven't had my caffeine yet.
The memory can train/the system POSTs, but the OS fails to load. With some more fiddling this may be feasible - I'm not sure. Excited to try!
Additionally: the IF frequency is 2100Mhz, trying to go much higher doesn't work out. If I could get 2200 it'd factor nicely into the memory frequency too, but alas - chip isn't up to it.
you could try 6400 on memory synced if 6600 boots with 4 dimms which is a very good result. 6600 is into the territory of needing luck with IO die and then also tuning all the voltages.
For the infinity fabric if you're trying to push it checking for performance regressions is a good idea as it delays and retransmits as a form of pseudo ecc when it detects an error. Same with memory if you're not using all custom timings as motherboards can set some very much not good timings if left on auto and sitting on the edge of stability
For the infinity fabric if you're trying to push it checking for performance regressions is a good idea as it delays and retransmits as a form of pseudo ecc when it detects an error. Same with memory if you're not using all custom timings as motherboards can set some very much not good timings if left on auto and sitting on the edge of stability
I appreciate the tips, will give it a try :) When speccing out this system/building it, I chased memory frequency as we can tell.
Seeing 1:1 is potentially within reach, perhaps dropping down to 6400 is worthwhile. Thanks again, will give it a whirl!
While I have chosen the EXPO profile, history has shown boards can't reliably read the (sub)timings... so I do it/apply those :P I try not to mess with those much, I don't really know the relationships between them.
Seeing 1:1 is potentially within reach, perhaps dropping down to 6400 is worthwhile. Thanks again, will give it a whirl!
While I have chosen the EXPO profile, history has shown boards can't reliably read the (sub)timings... so I do it/apply those :P I try not to mess with those much, I don't really know the relationships between them.
For the most part it's usually best to just go lowest that's stable, there are some "rules" online but most of them are just random things someone came up with. There are a couple JEDEC ones but the memory controller will adjust to work with those or it'd just not work.
A 1:2 sync usually starts being worth it above ~7200MTs but I don't know how viable that'd be with 4 dimms, at the very least it'd need tuning the various termination impedances
A 1:2 sync usually starts being worth it above ~7200MTs but I don't know how viable that'd be with 4 dimms, at the very least it'd need tuning the various termination impedances
Personally I don’t want maximum performance. I want maximum reliability, and minimum amount of time spent configuring the hardware.
If I were assembling a new computer tomorrow, I would probably get the 65W Ryzen 7 9700X with two 5600MT/s ECC unbuffered memory modules.
If I were assembling a new computer tomorrow, I would probably get the 65W Ryzen 7 9700X with two 5600MT/s ECC unbuffered memory modules.
Still, 4 modules should have more memory bandwidth than 2, even if the modules are themselves slower.
That would be the case if the CPU had four memory channels, but consumer platforms only have two. When you install four modules the bandwidth is shared between each pair, so on paper you end up with the same bandwidth as if you used two modules at the same clock, but running four modules puts additional strain on the CPUs IMC and demands better motherboard signal integrity so the clock speed usually has to be reduced to get it stable.
> Ryzen 7 9700X: 8 Zen 5 cores, 32 MB L3 cache, 65W TDP, available Aug 8
Hmmm max TDP listed at 65W that's lower than my 7700X which listed max TDP at 105W I think. I know people have been running the 7700X with big fat fanless CPU heatsinks.
So I take it that even if these TDP ratings are a bit of a lie, it's still not much and it should be possible to run a 9700X without any fan on the CPU heatsink?
Hmmm max TDP listed at 65W that's lower than my 7700X which listed max TDP at 105W I think. I know people have been running the 7700X with big fat fanless CPU heatsinks.
So I take it that even if these TDP ratings are a bit of a lie, it's still not much and it should be possible to run a 9700X without any fan on the CPU heatsink?
> DDR5-5600 when 2 modules are installed, DDR5-3600 when 4 modules are installed
Is this a new thing? I don't rememebr four sticks being slower than two on previous generations. Then again I haven't been paying too close attention these past few years.
Is this a new thing? I don't rememebr four sticks being slower than two on previous generations. Then again I haven't been paying too close attention these past few years.
New with DDR5, and it's the reason no-one should consider upgrading memory by adding additional sticks on DDR5 platform -- instead, if you need more memory, you should replace your modules. The additional two slots are basically a newbie trap.
Next gen will do away with this because it moves to CAMM2 and there will only be a single connector on consumer motherboards.
Next gen will do away with this because it moves to CAMM2 and there will only be a single connector on consumer motherboards.
I'm buying a server with 12 slots, should I go for 4x128 gb instead of 8x64 then?
You want to ideally have exactly one DIMM per channel. How many dimms this means depends on configuration, but modern EPYC platforms are 12-channel, so there is likely one slot per channel on your board.
You can run two DIMMs per channel without much downside. There's a dual-socketed server mobo out there that has 48 DIMM slots, making it possible to run 24TB of RAM in a 2U rack-mount
On current AMD processors, there is absolutely downside to running two DIMMs per channel. That's the discussion we're having here.
Is the server using udimms or rdimms? With rdimms the additional memory controller load from more dimms isn't nearly as much as with udimms
It's a DDR5 thing - four sticks have absolutely been slower on the Zen 4 too.
I've tried 4x32 GiB on a 7950X and I would get exactly that speed after several minutes (!) of POST.
Anything faster would not be stable, I ended up sending back half of it and settled with 64 GiB total, POST still takes 30s.
I have a similar setup and with 4x32 sticks the most I have been able get it 4800 with EXPO enabled. Is that your experience too?
It's not new (as 2 other commenters posted). See for example (DDR4 on X470): https://www.asrock.com/mb/AMD/X470%20Master%20SLI/index.asp#... and also (DDR4 on X570): https://www.asrock.com/mb/AMD/X570%20Steel%20Legend/index.as...
Scroll down to the tables listing UDIMM Memory Slot and you can see how different configurations of single rank vs. double rank play out with respective to what becomes the "max" frequency.
Scroll down to the tables listing UDIMM Memory Slot and you can see how different configurations of single rank vs. double rank play out with respective to what becomes the "max" frequency.
> They all support ECC memory, with the caveat that it requires motherboard support.
Is that different than the 7000 series? Basically: is this "ECC support all across the models" a 9000 thing or a DDR5 one?
P.S: you didn't list the 9700X in your summary (the one I'm eyeing on)
EDIT: ah no I see: you did list it but you made a typo and wrote 9600X twice.
Is that different than the 7000 series? Basically: is this "ECC support all across the models" a 9000 thing or a DDR5 one?
P.S: you didn't list the 9700X in your summary (the one I'm eyeing on)
EDIT: ah no I see: you did list it but you made a typo and wrote 9600X twice.
Thanks, corrected the typo
Typo: I think that second one should be 9700X, not 9600X again.
While these CPUs look great, I'm not digging the lack of PCIe lanes on the AM5 platform.
Still, might finally be time to upgrade my desktop.
Still, might finally be time to upgrade my desktop.
Are you coming from a HEDT platform? AM5 has 28 PCIe 5.0 lanes as opposed to 24 in AM4 and 20 in Intel Z790.
Yeah, was hoping for a bit more. But at least now there are AM5 motherboards with three PCIe x16 slots, so looking a lot more tempting.
The relatively high TDPs are putting me off Ryzen 9 a bit, even if you are getting 12-16 cores.
You always have the option to limit the TDP via the BIOS with nearly no performance hit (at least according to the benchmarks I made, and I also feel no difference when using my PC)
(Reposting a comment I made two weeks ago https://news.ycombinator.com/item?id=41043961)
In my BIOS I can activate a "ECO Mode", which lets me decide if I want to run my 7950x on full 170W TDP, 105W TDP or 60W TDP.
I benchmarked it, the difference between 170 and 105 is basically zero, and the difference to 60W is just a few percent of a performance hit, but way worth it, as it's ~0.3€/kWh over here.
(Reposting a comment I made two weeks ago https://news.ycombinator.com/item?id=41043961)
In my BIOS I can activate a "ECO Mode", which lets me decide if I want to run my 7950x on full 170W TDP, 105W TDP or 60W TDP.
I benchmarked it, the difference between 170 and 105 is basically zero, and the difference to 60W is just a few percent of a performance hit, but way worth it, as it's ~0.3€/kWh over here.
> the difference between 170 and 105 is basically zero
From tests I've read, if you would compile something or render or encode/transcode (on CPU) or similar, there is a measurable effect. It will take only little bit longer but consume noticeably less energy.
The other positive effect: reduced noise from fan or even no noticeable noise at all.
> 170W TDP, 105W TDP or 60W TDP
I'm curious why it is not possible to adjust these values in only few watt steps.
From tests I've read, if you would compile something or render or encode/transcode (on CPU) or similar, there is a measurable effect. It will take only little bit longer but consume noticeably less energy.
The other positive effect: reduced noise from fan or even no noticeable noise at all.
> 170W TDP, 105W TDP or 60W TDP
I'm curious why it is not possible to adjust these values in only few watt steps.
> From tests I've read, if you would compile something or render or encode/transcode (on CPU) or similar, there is a measurable effect. It will take only little bit longer but consume noticeably less energy.
One of my "benchmarks" included compiling a large rust project, I don't have the exact numbers, but in tge end it was such a tiny fraction (something like 5 seconds difference, with a compilation time of 2 minutes). Which is definitely not worth the ~70% increased power consumption.
> I'm curious why it is not possible to adjust these values in only few watt steps.
Don't know probably to keep it simple (and maybe nor to Fry your CPU if you swap your CPU to one which is rated for e.g. 65W TDP).
This "ECO Mode" has three settings:
"OFF" which means max TDP (in the case of my CPU 170W)
"105W",
"ON" which is 65W.
It also says in the description that "105W" mode only exists for CPUs with 170W TDP.
Also if you look at the other CPUs from AMD you'll see that they only sell CPUs with these TDP numbers and additionally 120W (the x3d variants), that's probably why they chose theese numbers.
One of my "benchmarks" included compiling a large rust project, I don't have the exact numbers, but in tge end it was such a tiny fraction (something like 5 seconds difference, with a compilation time of 2 minutes). Which is definitely not worth the ~70% increased power consumption.
> I'm curious why it is not possible to adjust these values in only few watt steps.
Don't know probably to keep it simple (and maybe nor to Fry your CPU if you swap your CPU to one which is rated for e.g. 65W TDP).
This "ECO Mode" has three settings:
"OFF" which means max TDP (in the case of my CPU 170W)
"105W",
"ON" which is 65W.
It also says in the description that "105W" mode only exists for CPUs with 170W TDP.
Also if you look at the other CPUs from AMD you'll see that they only sell CPUs with these TDP numbers and additionally 120W (the x3d variants), that's probably why they chose theese numbers.
You can always just change the power limits in the BIOS settings.
Thanks. So nice to see official ECC support come back to the non-PRO parts.
I'm rooting for AMD. Their 3D architecture was brilliant for desktops.
Re intel, I've followed this quite a bit. The intel problem is important enough to wipe out their stock, letting 18,000 ppl go and extending their warranty to two years. The PR side of things is truly awful. I wouldn't quote Toms hardware on this... This is a full account https://www.youtube.com/watch?v=b6vQlvefGxk
Re intel, I've followed this quite a bit. The intel problem is important enough to wipe out their stock, letting 18,000 ppl go and extending their warranty to two years. The PR side of things is truly awful. I wouldn't quote Toms hardware on this... This is a full account https://www.youtube.com/watch?v=b6vQlvefGxk
A lot of people root for AMD.
I heard recently the Internet in general loves AMD and gives the company a lot of good will even through their failures or their products are somewhat inferior. Maybe it's because they posed themselves as selling products for the people, more competitive prices, support for open source operating systems and so on. Or maybe just because they were the underdog in the industry.
I heard recently the Internet in general loves AMD and gives the company a lot of good will even through their failures or their products are somewhat inferior. Maybe it's because they posed themselves as selling products for the people, more competitive prices, support for open source operating systems and so on. Or maybe just because they were the underdog in the industry.
My issue with AMD is the difference between potential and having all the elements to realize that potential. It's on thing to have a great CPU (or GPU) design, but a company has to have all the platform around it working well and ideally at launch, plus developer support so end users can see the maximum benefit out of their purchase. One of the things I've noticed about the Ryzens is that it seems wise to wait a few months for a few AGESA updates to filter through, and another aspect is clamping down on motherboard manufacturers doing their own shenanigans when they want to boost benchmarks versus their competition
AMD CPUs are currently clearly superior to Intel, across the board.
AMD consumer GPUs are substantially inferior to NVidia, and can only sell at such low prices that AMD has trouble getting any margins from them. Hopefully they have better products next gen, NV really needs competition.
AMD consumer GPUs are substantially inferior to NVidia, and can only sell at such low prices that AMD has trouble getting any margins from them. Hopefully they have better products next gen, NV really needs competition.
All of this, but also a lot of it is stockholders hyping their bags. Like GameStop, but toned down a bit. It’s tough because it makes it a topic where I can’t take anyone at face value, and I have to default to distrust.
People crap on AMD GPUs all the time. The reason why people praise AMD is that their CPUs are extremely competitive, performance wise. The CPUs are not inferior at all.
It's the Intel's monopolistic behaviour for me personally.
Yep. I've been around long enough to have seen Intel repeatedly demonstrate that without anyone pushing them forward they're happy to rest on their laurels. Competition here is better for everyone.
Love gamersnexus humor. The coupon code on their video: THISISFINE
I really like that the CPUs from the last few rounds have a mini-GPU on board. This is a 16 core, 4.3GHz base processor, with pcie 5. It's going to work really nicely for misc software dev. The onboard GPU probably can't do gaming in satisfactory fashion but it'll drive a bunch of monitors just fine.
> it'll drive a bunch of monitors just fine.
I wish AMD would tell me how many monitors. Intel's Ark will tell me a 14700K's iGPU supports 4 monitors. AMD's specs for the 8700G supports 4 monitors [0]. But their new 9900X [1] and older 7900X do not mention how many displays are supported. It seems to be left off their GPUs too [2]
[0] https://www.amd.com/en/products/processors/desktops/ryzen/80...
[1] https://www.amd.com/en/products/processors/desktops/ryzen/90...
[2] https://www.amd.com/en/products/graphics/desktops/radeon/700...
I wish AMD would tell me how many monitors. Intel's Ark will tell me a 14700K's iGPU supports 4 monitors. AMD's specs for the 8700G supports 4 monitors [0]. But their new 9900X [1] and older 7900X do not mention how many displays are supported. It seems to be left off their GPUs too [2]
[0] https://www.amd.com/en/products/processors/desktops/ryzen/80...
[1] https://www.amd.com/en/products/processors/desktops/ryzen/90...
[2] https://www.amd.com/en/products/graphics/desktops/radeon/700...
It seems to be 4 displays; the marketing slides for the granite ridge explicitly list 4 displays https://www.tomshardware.com/pc-components/cpus/amd-dishes-m... . For Raphael it should be the same as the SOC is the same, I found some articles mentioning 4 displays but nothing directly from AMD quickly, they may have said it during a presentation
Isn't this related to mainboard/gpu board implementation? Maybe they do not like to get ahead of the vendors.
The rest of the system may limit the number of outputs in various ways, but the chip itself has its own maximum. It's important information which shouldn't be omitted.
And the external factors shouldn't matter here. In the same way they say 170W and not hide it because you may not have a good enough cooler to reach that power draw.
And the external factors shouldn't matter here. In the same way they say 170W and not hide it because you may not have a good enough cooler to reach that power draw.
With DisplayPort multi-stream transport, the number of ports aren't really relevant. It's not related to GPU power either; some RTX 4090s have five ports but Nvidia has a display limit of 4 displays too.
> I really like that the CPUs from the last few rounds have a mini-GPU on board.
Yeah I run a 7700X since it came out with its iGPU to drive my 38" monitor. Works flawlessly. It's also, arguably, a GPU without a fan. You still have a fan for the CPU: I did put a Noctua NH-12US on mine, which is silent in "powersave" CPU mode and barely audible in "ondemand" mode when the CPU is then working hard. Which is perfect, it's how I like my machine: a tiny audible feedback telling me it's cranking.
FWIW I run all but one of my virtual desktop in "powersave" mode, so the fan is never audible when I browse the web, edit PDFs, sort pictures or whatever and then one virtual desktop, the one where I code, puts the CPU in "ondemand" mode. I've configured my WM to automatically switch the CPU's mode depending on which virtual desktop I'm currently on. And if I happen to have one task that I really want to run in performance mode, I send it to my "coding" virtual desktop.
Long story short: for people who don't game and who don't run AI models on their main desktop computers, these kind of CPU with an iGPU are really nice.
Yeah I run a 7700X since it came out with its iGPU to drive my 38" monitor. Works flawlessly. It's also, arguably, a GPU without a fan. You still have a fan for the CPU: I did put a Noctua NH-12US on mine, which is silent in "powersave" CPU mode and barely audible in "ondemand" mode when the CPU is then working hard. Which is perfect, it's how I like my machine: a tiny audible feedback telling me it's cranking.
FWIW I run all but one of my virtual desktop in "powersave" mode, so the fan is never audible when I browse the web, edit PDFs, sort pictures or whatever and then one virtual desktop, the one where I code, puts the CPU in "ondemand" mode. I've configured my WM to automatically switch the CPU's mode depending on which virtual desktop I'm currently on. And if I happen to have one task that I really want to run in performance mode, I send it to my "coding" virtual desktop.
Long story short: for people who don't game and who don't run AI models on their main desktop computers, these kind of CPU with an iGPU are really nice.
> I've configured my WM to automatically switch the CPU's mode depending on which virtual desktop I'm currently on.
What WM are you using? Some tilling WM that's scriptable or a major one like KWin?
I want to script terminal behaviour based on virtual workspace number (or open windows, preferably). Basically launch in with profiles/commandline arguments; and not sure how I'd do that in a wayland compatible way.
What WM are you using? Some tilling WM that's scriptable or a major one like KWin?
I want to script terminal behaviour based on virtual workspace number (or open windows, preferably). Basically launch in with profiles/commandline arguments; and not sure how I'd do that in a wayland compatible way.
While it’s probably not perfect for everyone, I know hyprland exposes the events you’d need for this, accessible via IPC.
I wish it was a tiny bit better, or it's just memory bandwidth issues, but when I was running purely from igpu a 4k video stream made animations etc. choppy
This might depend on what / how you're doing this?
I daily drive an older zen3 mobile part, with one external 4k screen and the internal display panel.
Never had any kind of issue playing hw decoded 4k videos on Linux with picom composer (doing some shadows / blur on random windows). Both windowed and fullscreen.
It also seems to work fine on windows, though I don't use that too often, so I may have missed something.
I daily drive an older zen3 mobile part, with one external 4k screen and the internal display panel.
Never had any kind of issue playing hw decoded 4k videos on Linux with picom composer (doing some shadows / blur on random windows). Both windowed and fullscreen.
It also seems to work fine on windows, though I don't use that too often, so I may have missed something.
The 4k display being at 144Hz may not be helping things if the bandwidth is the issue.
The igpu may also be a bit broken as it works fine with passthrough from my dgpu but it was not having a good time running everything off of it with monitors plugged directly into its ports. Browser hw acceleration sometimes caused what seemed like memory artifacts https://u.numerlor.me/PTZj and driver crashes weren't uncommon before I gave up on it. But I don't know if that'd affect performance, and RAM is perfectly stable as that's overclocked and tested rigorously on its own
The igpu may also be a bit broken as it works fine with passthrough from my dgpu but it was not having a good time running everything off of it with monitors plugged directly into its ports. Browser hw acceleration sometimes caused what seemed like memory artifacts https://u.numerlor.me/PTZj and driver crashes weren't uncommon before I gave up on it. But I don't know if that'd affect performance, and RAM is perfectly stable as that's overclocked and tested rigorously on its own
The only machine I've ever had with this GPU setup was a 2013 MBP, which ran perfectly.
But yeah, I've heard horror stories with newer multi-gpu laptops, especially on Linux.
So, since I've been happy with the 4k performance of iGPUs ever since the hd530 of the 6th gen core, I've specifically looked for iGPU-only laptops.
But yeah, I've heard horror stories with newer multi-gpu laptops, especially on Linux.
So, since I've been happy with the 4k performance of iGPUs ever since the hd530 of the 6th gen core, I've specifically looked for iGPU-only laptops.
This was on my desktop with a 7600x and a 7900xtx, I never managed to properly route the dGPU when I have everything plugged into iGPU, seemed like a driver issue where most programs never managed to actually attach to the GPU and got stuck in some driver hell as they were also unkillable.
Only managed to get compute and a Furmark vulkan test working otherwise it only works with at least one screen plugged directly into the dGPU
Huh. I didn't even know this was a thing on desktops. Is this supposed to work with things like thunderbolt displays?
Is the Intel "silicon degradation" problem real? Is there any chance this is also around the corner for AMD?
To answer the first question:
Yes and Intel has confirmed it. It should be fixed now but too high voltages can damage CPUs permanently so it probably means a lot of warrabty claims
See https://www.tomshardware.com/pc-components/cpus/intel-cpu-in...
The problem is unlikely to be around the corner for AMD since they do not have to supply as much voltage to their chips as as Intel to keep up since they are - at least when it comes to x86 chips - currently leading in performance and power efficiency
Yes and Intel has confirmed it. It should be fixed now but too high voltages can damage CPUs permanently so it probably means a lot of warrabty claims
See https://www.tomshardware.com/pc-components/cpus/intel-cpu-in...
The problem is unlikely to be around the corner for AMD since they do not have to supply as much voltage to their chips as as Intel to keep up since they are - at least when it comes to x86 chips - currently leading in performance and power efficiency
It is not fixed as Intel hasn't published their microcode update, which is lowering the voltages, yet.
Which will likely affect performance as well, right? As I understood it, they were too ambitious with pushing performance using higher voltages, which now needs to be reduced using a microcode update?
Either performance or stability/reliability. Or even both.
At higher frequencies (during a frequency boost phase for example) signal quality in the digital signals degrades, because the "stable 1" or "stable 0" plateau is shortened and the "maybe 1 or 0" phase in between stays the same. So a signal that is supposed to be as rectangular as possible gets smushed down towards a sinus, and then smushed even further towards lower amplitudes.
One measure against this is of course better (faster) transistors, such that the "maybe" phase is shorter, but that only works by replacing the hardware. The other measure, which you can do during runtime, is to increase the signal amplitude by increasing the voltage. Then even a degraded signal close to the transistors' maximum switching frequency gets over the "stable 1" and "stable 0" thresholds fast enough.
With a lower supply voltage you can thus not clock the CPU as high as before which is important in boost phases during high load. Which would decrease peak performance in standard desktop and server workloads, and decrease overall performance in compute-intensive workloads. Or if you still clock it as high as before, signal quality will be lessened, increasing the probability of bit errors, lessening system stability and reliability of results.
Which direction intel will pick for this firmware upgrade, degraded-performance, degraded-stability or degraded-both, I don't know, I guess we'll see.
At higher frequencies (during a frequency boost phase for example) signal quality in the digital signals degrades, because the "stable 1" or "stable 0" plateau is shortened and the "maybe 1 or 0" phase in between stays the same. So a signal that is supposed to be as rectangular as possible gets smushed down towards a sinus, and then smushed even further towards lower amplitudes.
One measure against this is of course better (faster) transistors, such that the "maybe" phase is shorter, but that only works by replacing the hardware. The other measure, which you can do during runtime, is to increase the signal amplitude by increasing the voltage. Then even a degraded signal close to the transistors' maximum switching frequency gets over the "stable 1" and "stable 0" thresholds fast enough.
With a lower supply voltage you can thus not clock the CPU as high as before which is important in boost phases during high load. Which would decrease peak performance in standard desktop and server workloads, and decrease overall performance in compute-intensive workloads. Or if you still clock it as high as before, signal quality will be lessened, increasing the probability of bit errors, lessening system stability and reliability of results.
Which direction intel will pick for this firmware upgrade, degraded-performance, degraded-stability or degraded-both, I don't know, I guess we'll see.
Maybe. I don't think anyone had information yet regarding the details. The extra power may have been just pure waste, rather than a performance improvement.
Unless I missed a real investigation posted somewhere?
Unless I missed a real investigation posted somewhere?
> for AMD since they do not have to supply as much voltage to their chips
... unless you're using EXPO/XMP. I would really recommend people who run automatically overclocked memory taking a closer look at how much voltage the motherboard is pushing into your CPU, especially into the SoC. Some motherboards just push the voltages to the absolute permitted maximum; ASUS is particularly bad at this. I run lower RAM frequencies than my system is capable of because I haven't seen much performance improvement above 5600 "MHz", and it can be made to run at almost stock voltages.
It definitely causes faster silicon degradation; the question is how fast it will kill the processor. Both Intel and AMD have shown us that it can happen very quickly, not in the course of several decades as we've assumed for the longest time.
... unless you're using EXPO/XMP. I would really recommend people who run automatically overclocked memory taking a closer look at how much voltage the motherboard is pushing into your CPU, especially into the SoC. Some motherboards just push the voltages to the absolute permitted maximum; ASUS is particularly bad at this. I run lower RAM frequencies than my system is capable of because I haven't seen much performance improvement above 5600 "MHz", and it can be made to run at almost stock voltages.
It definitely causes faster silicon degradation; the question is how fast it will kill the processor. Both Intel and AMD have shown us that it can happen very quickly, not in the course of several decades as we've assumed for the longest time.
It seems new to me that the safety margins are so narrow that a microcode bug can result in voltages that physically damage the CPU. Assuming that's mostly driven by the current small feature sizes and high clock rates, that would mean everyone has this risk.
Toasting CPUs via overclocking and overvolting has been possible for a long time. Back in the day they didn't even have thermal throttling.
It just used to be jumpers and dip switches, then the BIOS before some of it moved into the CPU itself.
It just used to be jumpers and dip switches, then the BIOS before some of it moved into the CPU itself.
I'd consider that to be one more driver (bios/software selectable jumpers) in addition to increasingly small features and high clock rates. Rather than something that refutes the situation is new.
It's pretty unlikely that it's around the corner. Intel uses their own foundry and designs and AMD uses their own designs with TSMC as the foundry. If it's a foundry issue then Apple would be in hot water right now since they use the cutting edge. So I think AMD is in the clear.
It’s not a foundry issue, it’s a “how far can we push voltage for better performance to compensate for older architectures” problem. That’s why it can be resolved using a microcode update.
Via oxidation is a foundry issue.
>Intel also confirmed a rumored issue with via oxidation in its 7nm node, but said those issues were corrected in 2023 and didn't contribute to the failures.
Do you trust Intel to not lie about a widespread unfixable problem?
https://www.tomshardware.com/pc-components/cpus/intel-finall...
>Intel also confirmed a rumored issue with via oxidation in its 7nm node, but said those issues were corrected in 2023 and didn't contribute to the failures.
Do you trust Intel to not lie about a widespread unfixable problem?
https://www.tomshardware.com/pc-components/cpus/intel-finall...
I trust them not to make provably false statements, yeah.
Walking the line of "well, we didn't say..." is one thing but if they say the oxidation issue was ended in may 2023 then I don't see a reason to doubt that.
Walking the line of "well, we didn't say..." is one thing but if they say the oxidation issue was ended in may 2023 then I don't see a reason to doubt that.
And the 13th gen CPUs sold between October 2022 and May 2023? "Do not contribute to failures" can be technically not a lie because so much is omitted by not specifying which failures.
What about them? There’s no reason to believe oxidation defects during manufacture is a cause of those failures.
This is a complex problem with several interacting causes and failure modes, and assuming that oxidation is the cause of all of them (that there exist chips not during the oxidation defect window that failed, therefore the oxidation defect window must be larger) is circular reasoning.
This is a complex problem with several interacting causes and failure modes, and assuming that oxidation is the cause of all of them (that there exist chips not during the oxidation defect window that failed, therefore the oxidation defect window must be larger) is circular reasoning.
How far you can push the voltage is definitely foundry technology dependent. If Intel foundry internally specced a voltage that actually damages a chip I would say it IS a foundry issue.
Silicon chips degrading under high temperature and voltage is very well known. "Overclocking" was (is?) the game of deliberately turning CPU frequency up to get better performance, and turning voltage up to get stable performance, while trying to keep the chip cool enough for this to broadly work out. If you push the envelope a bit hard the stable overclock becomes unstable over time and turning down the voltage and frequency can make it stable again. That behaviour is pretty well documented. Happens to memory too.
When you draw more current - i.e. make the chip do work - the voltage drops. "Load line calibration" was introduced as a way to crank the applied voltage up as the chip worked harder to offset this. Alternatively phrased, if you set the voltage so that it works well under load, it's kind of set too high for idle, so LLC gives a way to decrease idle voltage.
There was some discussion at the time about the wisdom of this. The voltage rings when you change it (overshoots), and it's not that clear whether high voltage at idle matters very much (since the chip is cooler then). I remember reading a bunch of articles, then some materials science books, then turning off LLC in the bios as probably a misfeature.
The current Intel problem is they've built a reputation on being the fastest and the most reliable CPU. Then they lost fastest to chiplets at AMD, so adjusted to "fastest single core". They really didn't want to lose that too, and CPU's run faster when you push more voltage through them, so that's what they did.
---
"Intel silicon degradation" currently refers to people noticing that desktop processors from the last couple of generations were degrading, despite not being overclocked. Some finger pointing there, Intel encourages motherboards to set fairly aggressive out of the box controls because it looks good under review. Also some more recent reports that mobile chips have the same behaviour. At a guess, the desktop ones have the voltage too high and the mobile ones are running at much higher temperature. Some reports are mentioning LLC which would be personally satisfying if it turns out to be the root cause.
I'm waiting to see if reports of degradation start to emerge on the Xeon line. Intel is claiming there's no risk of that but they're made out of the same magic sand and under really heavy competitive pressure from the Epyc chips. They've also got rather high power consumption relative to previous generations.
As to whether the same thing will hit AMD - probably not. If you take one of their chips and push the voltage up, yeah, it'll have the same degradation over time. But they're not running out of the box at the edge of feasible to try to cope with the competition because chiplets are winning the day. Also AMD doesn't have the gilded reputation of Intel to rely upon if reliability comes under question. The incentives are pointing in the opposite direction.
When you draw more current - i.e. make the chip do work - the voltage drops. "Load line calibration" was introduced as a way to crank the applied voltage up as the chip worked harder to offset this. Alternatively phrased, if you set the voltage so that it works well under load, it's kind of set too high for idle, so LLC gives a way to decrease idle voltage.
There was some discussion at the time about the wisdom of this. The voltage rings when you change it (overshoots), and it's not that clear whether high voltage at idle matters very much (since the chip is cooler then). I remember reading a bunch of articles, then some materials science books, then turning off LLC in the bios as probably a misfeature.
The current Intel problem is they've built a reputation on being the fastest and the most reliable CPU. Then they lost fastest to chiplets at AMD, so adjusted to "fastest single core". They really didn't want to lose that too, and CPU's run faster when you push more voltage through them, so that's what they did.
---
"Intel silicon degradation" currently refers to people noticing that desktop processors from the last couple of generations were degrading, despite not being overclocked. Some finger pointing there, Intel encourages motherboards to set fairly aggressive out of the box controls because it looks good under review. Also some more recent reports that mobile chips have the same behaviour. At a guess, the desktop ones have the voltage too high and the mobile ones are running at much higher temperature. Some reports are mentioning LLC which would be personally satisfying if it turns out to be the root cause.
I'm waiting to see if reports of degradation start to emerge on the Xeon line. Intel is claiming there's no risk of that but they're made out of the same magic sand and under really heavy competitive pressure from the Epyc chips. They've also got rather high power consumption relative to previous generations.
As to whether the same thing will hit AMD - probably not. If you take one of their chips and push the voltage up, yeah, it'll have the same degradation over time. But they're not running out of the box at the edge of feasible to try to cope with the competition because chiplets are winning the day. Also AMD doesn't have the gilded reputation of Intel to rely upon if reliability comes under question. The incentives are pointing in the opposite direction.
MSRP comparison with 7000 series:
7600: $229, 9600X: $279
7700: $329, 9700X: $359
7900X: $549, 9900X: $499
7950X: $699, 9950X: $649
(using non-X variant for 7600 & 7700 since they have same 65W TDP)
7600: $229, 9600X: $279
7700: $329, 9700X: $359
7900X: $549, 9900X: $499
7950X: $699, 9950X: $649
(using non-X variant for 7600 & 7700 since they have same 65W TDP)
Interesting, the R9's are cheaper than the previous gen. Hmmmm, that makes an upgrade tempting...
Some speculate that Intel's recent failure is partially to blame for AMD's price bump.
The prices all decreased if you compare X variant to X variant.
The 7600X launched at $299, and the 7700X launched at $399.
The 7600X launched at $299, and the 7700X launched at $399.
my bad, indeed. I thought my parent commenter had launch price for both sides.
The speculation on price bump did indeed happen when intel problem was widespread recently.
The speculation on price bump did indeed happen when intel problem was widespread recently.
I wonder if the 9700 end up being better for gaming due to the way lower tdp while already having 8 cores, the requirements for cooling drop substantially so overclocking could be used to boost single core which could be more useful
Is there a reason why a few AMD motherboards have Thunderbolt 4 on them but most do not? I believe I’ve only seen it on the workstation class boards, I could be wrong though. Is there a reason why it’s not more broadly available on AMD mobos?
Thunderbolt isn't an open standard and Intel holds the rights to it.
The next generation of motherboards is supposed to make USB 4 standard. Those are coming some time after the CPU release though.
Just wondering - What would thunderbolt be used for on a desktop? External SSD transfers? I am shopping for a new gaming PC build and didn't think to consider it. I have it on my laptop for one-cable docking.
Ive used it to swap my GPU between my desktop and laptop. My desktop is more of a server, and my laptop is more of a workstation. Having the option is nice.
Anyone know how much GPU you need to bottleneck at a 9600x? Will it take a full 4090 without a performance impact? Or does it start to be a problem with a 4080 or 4070super.
Ryzen 5 9600X: 6 Zen 5 cores, 32 MB L3 cache, 65W TDP, available Aug 8
Ryzen 7 9700X: 8 Zen 5 cores, 32 MB L3 cache, 65W TDP, available Aug 8
Ryzen 9 9900X: 12 Zen 5 cores, 64 MB L3 cache, 120W TDP, available Aug 15
Ryzen 9 9950X: 16 Zen 5 cores, 64 MB L3 cache, 170W TDP, available Aug 15
All models support dual-channel DDR5, the speed is DDR5-5600 when 2 modules are installed, DDR5-3600 when 4 modules are installed. They all support ECC memory, with the caveat that it requires motherboard support.