Thanks! This information is not on any of the 7x00X product pages. I suppose the max DP version mostly answers my question.I'm sure someone could easily find it on AMD's site... but I'm lazy this morning.
Awesome. 5800X3Ds are hovering around $315 in the US so that’s a great deal up yonder. The 5700X3Ds on the other hand are quite affordable here at about $245.PSA for Canadians, the 5800X3D has a coupon that takes it down to CA$320 at canadacomputers, or CA$20 less than a 5700X3D. The coupon code is pre-filled on the shopping cart, and clicking on "apply coupon" will reduce the price. In-store pick-up only though.
In related news, I expect to put a 5600X on the agora soon
Yeah, it's the equivalent of US$236, so I didn't feel too bad about getting a Thermalright Peerless Assassin to go with it. Just have to wait for that to arrive.Awesome. 5800X3Ds are hovering around $315 in the US so that’s a great deal up yonder. The 5700X3Ds on the other hand are quite affordable here at about $245.
Congrats on the upgrade
"Inexpensive" and "future proof" don't often go together in personal computers. More often than not, "future proofing" is an upselling strategy from the marketing department.
The video currently has 59 thousand views. I think the information will trickle up sooner rather than later.
Hardware Unboxed's testing platform for many of their 5800X3D videos was a B350 ATX board...
If there is a point you want to make then please spell it out.
I think he's pointing out that in the prior generation, even the cheapest piece-of-shit motherboards would handle the fastest CPU on the socket just fine.
Even if they did, the jump probably wouldn't be huge for most people. Even 32 threads is getting unrealistic for most workloads. Algorithms that go that widely multithreaded are often being offloaded to the GPU anyway.
It's not that simple. GPUs have stricter constraints on what they can and cannot run efficiently. Multithreading on manycore CPUs isn't trivial, but generally easier than GPGPU programming. The toolchain for multithrading has improved a lot in recent times, catalysed by the availability of affordable 16+ threaded machines.
Rendering, encoding, running VMs, running multiple game servers, generating ML models, compiling the Linux kernel to name a few.
The number of algorithms that need wide multithreading but can't run on GPUs is a very small fraction of what's out there.
The Zen 3 5800X was a massive upgrade from a 4790K, so going all the way to Zen 5 should tingle your little toes.
High performance and scientific computing, a range of algorithms for genetics and bioinformatics scale really well with cpu cores. Photography also benefits when exporting and importing images.
Sure, but you've moved way out of the mainstream there. By and large, that's serious professional use territory.
One prominent example these days is raytracing. This light simulation has ungodly amounts of parallelism, but it is very difficult to map it to the lock-step of SIMD hardware in a scalable way. So 4-wide SIMD is okay-ish, because we can put a 3-vector of a single ray into a SIMD register. But 32-wide or 64-wide of a GPU means tracing several rays together ... only that those rays usually veer off into entirely different parts of the scene.
Programming would certainly be simpler if we could get the IPC we're used to from a single 40 GHz core instead of eight 5 GHz cores. And the user experience would be better, too, because every single piece of software would benefit from that.
I think the steam survey recently saw the peak move from four cores to six cores, but the average core number was already closer to eight. At the same time, enthusiast PC reviewers have begun to talk of six cores as entry level, and are recommending eight cores for a new gaming rig.
