A 2800x wouldnt/will not make much difference. It would still be behind in pubg (that i can't find on the list of games on the ppt) and old stuff.Perhaps the explanation is quite simple - they expect a minor respin to yield them enough improvement for a clock bump later on, so can launch a 2800X in 6 months time.
into the 2950x and it will be a tough processor that beats all the sklx stuff.
What are the consumer workloads that you are talking about? Just curious. I never thought that a person would buy 7960X class machine for consumer class workload. In my mind, consumer workload means web browsing, document editing, gaming (not 120Hz competitive gaming), etc. So, I like to hear what you consider consumer workloads.As a 7960X owner I might be biased, but IMO you cannot compare Threadripper with any of the SKL-X SKUs. Threadripper is a 2P solution and that causes some limitations, or is otherwise generally unoptimal for consumer workloads.
No, it's definitely that they are holding back.
Why not ? its not a 2P solution, thats EPYC. AMD even says that your 7960X is the competition for the 1950X, and at $1565 vs $900 (current prices.)As a 7960X owner I might be biased, but IMO you cannot compare Threadripper with any of the SKL-X SKUs. Threadripper is a 2P solution and that causes some limitations, or is otherwise generally unoptimal for consumer workloads.
As a 7960X owner I might be biased, but IMO you cannot compare Threadripper with any of the SKL-X SKUs. Threadripper is a 2P solution and that causes some limitations, or is otherwise generally unoptimal for consumer workloads.
Why not ? its not a 2P solution, thats EPYC. AMD even says that your 7960X is the competition for the 1950X, and at $1565 vs $900 (current prices.)
If you set it to gaming mode, you one then have one numa node I am pretty sure..... It does BOTH.I think he means the NUMA-Nodes and their depending problems with them. SKL-X has only 1 NUMA-Node.
If you set it to gaming mode, you one then have one numa node I am pretty sure..... It does BOTH.
Non-consumer workloads shouldn't suffer much from two different NUMA Nodes because they are parallelized carefully to avoid inter-thread dependencies as much as possible. If however windows assigns consumer tasks with less than 8 threads to cores across two different NUMA nodes that would just be a bad scheduler implementation of Microsoft.
I don't really get why Windows moves tasks across cores any way. Sure that distributes heat better, but it also causes a lot of unnecessary cache faults and context switches. They should rather keep tasks at a core and don't interrupt it if possible.
“Win7 keeps all physical cores awake, and parks SMT cores. Win10 keeps one physical and one logical core away (Core0+1), then parks the rest as often as possible. This change alone is what’s responsible for the cases where Win7 was faster than Win10 gaming performance, not the scheduler as the community thought.”
Skl x comes factory oc out the box with tdp numbers beeing wrong.As a 7960X owner I might be biased, but IMO you cannot compare Threadripper with any of the SKL-X SKUs. Threadripper is a 2P solution and that causes some limitations, or is otherwise generally unoptimal for consumer workloads.
Why not ? its not a 2P solution, thats EPYC. AMD even says that your 7960X is the competition for the 1950X, and at $1565 vs $900 (current prices.)
Skl x comes factory oc out the box with tdp numbers beeing wrong.
While that might technically be the case, the statement above is still untrue.
Just like any other CPU from AMD or Intel, SKL-X obeys it's set power limits.
The issue isn't that Intel would be shipping them with a wrong TDP rating, because they aren't.
The issue is that the motherboard ODMs are altering the default power limits (PL1, PL2) and their durations (Tau) automatically.
For 165W SKL-X SKUs the power limits default and should remain at 165W (PL1) and 206W (PL2). However on most systems, including my own
the power limits defaulted to 4096W (automatically set by the ODM). Obviously any power limited chip, which has it's power limiters disabled will exceed it's advertized TDP.
I've measured the actual power consumption of the 7960X using DCR, with the power limits and times correctly configured and the CPU obeys them just as it should.
Same exact thing is done by the ODMs on Z370 platform as well.
Are there any indications that the new zen+ motherboards will have onboard graphics?
Threadripper IS a 2P solution. The only difference to conventional MP solutions is that both of the processors reside on the same substrate.
If the workloads are NUMA aware, then obviously everything is fine.
However if, or rather when the workloads aren't NUMA aware you'll should be using UMA mode instead.
In UMA mode the latency is pretty atrocious since some of the memory accesses always have to cross the die boundary (through GMI), causing the latency to increase even further beyond the already high base level.
Threadripper is an excellent solution for certain, very specific workloads especially when considering it's price.
However, personally I went with a 7960X instead because I wanted something which is a true single CPU solution and has better performance in >= 256-bit workloads.
And that's after the fact that I have 1950X and a board for it sitting in the shelf, so didn't make my choice based solely on assumptions either.
Anyways, it would be nice if you tell us some concrete examples where TR isnt a good solution, I mean concrete HEDT apps.
A 2800x wouldnt/will not make much difference. It would still be behind in pubg (that i can't find on the list of games on the ppt) and old stuff.
Instead put the best stuff aside and into the 2950x and it will be a tough processor that beats all the sklx stuff.
Which workloads you qualify as such?
256-bit workloads aren't too specific IMO, as all of the modern video encoders (X264, X265, VP9, AV1) implement AVX2 and some even AVX-512.
Among the music production software, and even with something non-exotic as Blender (with the physics engine implemented) you can see the further amplified effect of the high latency from TRs 2P config.
Obviously most software intended to be run on server grade hardware support NUMA anyway.
If NUMA is supported then obviously the latency isn't any more of an issue than it is on desktop Ryzens.
Well I've not checked lately, but when TR was released, it perfomed pretty well in video encoding tasks. SKL-X was only faster with x265 codec if I remember correctly. About Blender it peformed pretty well too, dont know about music production though
Behind yes, but if it means AMD have another unit they can charge a higher price for then it makes financial sense for them.
[There is a world beyond the epeen contest...]
One can say x265 is the important thing today and h264 less so. Time works for skl here.Well I've not checked lately, but when TR was released, it perfomed pretty well in video encoding tasks. SKL-X was only faster with x265 codec if I remember correctly. About Blender it peformed pretty well too, dont know about music production though