Stereodude
Not really a
Well that's disappointing. 
Ryzen
- Thread starter time
- Start date
sechs
Storage? I am Storage!
The same chiplets are used in Ryzen, Threadripper, and Epyc. I think that demand is just much, much higher than expected.
Considering TSMC's recent announcement concerning lead times on new chips, AMD could be in a crunch.
Considering TSMC's recent announcement concerning lead times on new chips, AMD could be in a crunch.
Stereodude
Not really a
But why what?Sure, but why?
Something is afoot.
Stereodude
Not really a
What's "this"?
sechs
Storage? I am Storage!
The new iPhones are often cited. Their chip is also produced by TSMC on 7nm.Sure, but why?
Something is afoot.
Between Apple and AMD, that's a lot of silicon.
Stereodude
Not really a
Does it really matter? You may never know the exact reasons why they delayed one CPU's launch.So the issue is simply a supply chain clusterf***?
Not really. I suppose now it would be best to wait and see how many GHz can be produced with their HEDT CPUs that have a good number of PICe lanes. I keep thinking about that Windows 10 and will need it sooner rather than later.
Stereodude
Not really a
I'm less concerned about raw gHz than I am actual performance. Chasing ultra-high gHz just burns power and clobbers efficiency.
Stereodude
Not really a
How far into the future are you considering? Intel is claiming a big IPC gain with their latest cores that would make them roughly on par with the Zen 2 cores (for IPC), but so far those are all on Intel's 10nm which clocks slower than their 14nm process (so far).
sechs
Storage? I am Storage!
I get the (unfounded) feeling that Threadripper will get better chiplets. The AM4 parts with two chiplets have a good one and a not as good one.
I think that Intel has spun out on HEDT. All they can do is push clocks at 14nm and hope that their 10nm chips can catch up. I don't think that they're even bothering with a new chipset for the next round.
I think that Intel has spun out on HEDT. All they can do is push clocks at 14nm and hope that their 10nm chips can catch up. I don't think that they're even bothering with a new chipset for the next round.
My time is running out. I cannot get the Sony software to run (though it supposedly does work in Win 7) and I'm sure that Nikon and Canon updates will no longer support 7 RSN.
Stereodude
Not really a
Apparently not completely spun out... https://www.anandtech.com/show/14925/intel-cascade-lakex-for-hedt-18-cores-for-under-1000
They're still 14nm though.
sechs
Storage? I am Storage!
Doesn't AMD already offer more than 18 cores for less than $1000?
This looks kind of like Intel trying to fight AMD's mainstream desktop with HEDT.
Chewy509
Wotty wot wot.
Ryzen Pro 3000 released...
https://arstechnica.com/gadgets/201...-desktop-cpus-will-offer-full-ram-encryption/
Unfortunately, currently only available to major OEMs.
https://arstechnica.com/gadgets/201...-desktop-cpus-will-offer-full-ram-encryption/
Unfortunately, currently only available to major OEMs.
Chewy509
Wotty wot wot.
Yes... https://www.vusec.net/projects/netcat/Are people hacking into the RAM now?
Stereodude
Not really a
Not with Zen 2 cores, at least not yet. They'll have 16 for $749 once they launch the 3950X. The next gen Zen 2 based Threadripper looks like it may start at 24 cores. No idea if they would price that at $999 or there will be variants with less than 24 cores that they could put at that price point.
It's just too much BS being connected to the internet. I don't want to deal with all that.
I would hope that some users want PCIe lanes and don't need 24 cores, but if there is a cheaply made CPU, then all bots are off.
sechs
Storage? I am Storage!
Well, Intel isn't doing it with Ice Lake, either.
Threadripper 1000 and 2000 chips have a much more compelling price/performance.
Stereodude
Not really a
Depending on your specific workload.
I think it depends more on the specific worker and percentage of time saved with a better CPU.
You must take into account the level or workers using them, e.g., GS6, GS9, GS12, higher? What are the fully burdened costs?
Newtun
Storage is nice, especially if it doesn't rotate
https://arstechnica.com/gadgets/2019/10/how-a-months-old-amd-microcode-bug-destroyed-my-weekend
I'd been "window shopping" for x570 mobos, and maybe that article explains why I saw a significant number of x570 purchasers that had problems getting their builds to boot, or had other inexplicable reliability issues.
I'd been "window shopping" for x570 mobos, and maybe that article explains why I saw a significant number of x570 purchasers that had problems getting their builds to boot, or had other inexplicable reliability issues.
Newtun
Storage is nice, especially if it doesn't rotate
I'm not sure what you mean, but there can't be too much of a risk if you can't boot your Ryzen 3000 PC
. (Apparently, this bug has been around for months [and somewhere I read that it's been years, but I'm not sure of that].)ExtremeTech.com said:
Aside from that issue with Destiny, it seems that Linux uses RDRAND in its boot-up cycle, maybe for virtual-memory address conflict avoidance or similar.
Editing to add some new references (The Linux problem seems to be in systemd, "the systemd "System and Service Manager", an init system used to bootstrap user space and manage user processes. - Wikipedia"):
Last edited:
Newtun
Storage is nice, especially if it doesn't rotate
Well, that would increase the pressure on AMD to correct the problem.
Apparently, AMD has provided a fix to motherboard manufacturers, for them to include in their BIOS updates. IIRC, this would correct the microcode on the Ryzen CPU itself, to make RDRAND produce varying "random" numbers, instead of the same 0xFFFFFFFFFFFFFFFF value every time it's called.
I think I saw some software fixes for the issue for Destiny 2 and systemd, but I'm not sure of the status of the systemd updates in the various Linux distros, and of course, those would only be "work-arounds" for the hardware problem.
Apparently, AMD has provided a fix to motherboard manufacturers, for them to include in their BIOS updates. IIRC, this would correct the microcode on the Ryzen CPU itself, to make RDRAND produce varying "random" numbers, instead of the same 0xFFFFFFFFFFFFFFFF value every time it's called.
I think I saw some software fixes for the issue for Destiny 2 and systemd, but I'm not sure of the status of the systemd updates in the various Linux distros, and of course, those would only be "work-arounds" for the hardware problem.
Stereodude
Not really a
More announcements and details: https://www.anandtech.com/show/15062/amds-2019-fall-update
November 25th is the release day for the 3950X, the 3960X, and 3970X.
November 25th is the release day for the 3950X, the 3960X, and 3970X.
Stereodude
Not really a
Who cares? Your power company will sell you the extra power you "need". Additionally, I still don't know why so you're hung up on PCIe lanes that you seem to have no need for.There is quite a gap between the 105W 3950X and 280W 3960X. 280W is too much power just to get a decent number of PCIe lanes.
Stereodude
Not really a
Well, they are more efficient than Intel. Their TDP ratings are also more honest. So a comparable HEDT Intel solution will use even more power under load and have worse performance.The tower is normally only a half meter away from me, so heat is an issue especially in the summer.
So you should be used to it.The electric company congratulated me for using significantly more power last year than similar homes.