Depending on the boost clocks you’ll get, that’ll be 3,7Ghz to 4,5Ghz on stock Ryzen 1700 vs. a 7700K. If the IPC on the Ryzen is a bit lower, that’s maybe 25%. That’s somewhat noticable, but not life-changing.
On the other hand, if you’re gaming and you don’t hit 60FPS because of that difference, that can be a bummer. Most gaming benchmarks didn’t show such a difference, but games requiring lots of draw calls (like ARMA) might.
I think for the average Blender user doing CPU renders, Ryzen is well worth the tradeoff.
As for the AVX2 performance penalty: Almost nobody uses (or will use) AVX2 to any significant degree, so cutting corners there makes total sense. Cycles uses it very sparingly. Even with renderers based on Embree, where Intel does lot of AVX optimization, the difference doesn’t seem substantial:
Boy, the twists in terms of what chip to have in a new machine keep on coming.
Turns out you can’t overclock the Kaby Lake at all, because people are reporting heat issues even at stock clocks (so Intel is saying not to overclock the chip)
Meanwhile, AMD Ryzen is getting a new patch that dramatically expands its memory-handling abilities (so putting RAM into the machine is now about as plug-n-play as Intel).
I’m comparing the (stock) boost clocks of the Ryzen 1700 (3.7Ghz) vs. the 7700K (4.5Ghz). At a comparable IPC, that’s a single core advantage of at most 21% for the 7700K. Assuming that the 7700K has slightly better IPC, a 25% advantage seems like a reasonable estimate.
Agreed with you BeerBaron. Luckily when dealing with Blender, more and more aspects are multi-threaded, as such, singe core boost is less and less of a factor. So most of that advantage is vanishing. Again in multithreaded tasks. gaming is still a mixed bag, but more on Intel side, there is no denying that.
Still, I was extremely dissapointed by Intel’s response to CPU overheating issues (even at stock)… which was… don’t overclock the K cpus.
If only AMD could get their speeds up a bit.
Either way, these days, even more so the CPU, GPU’s are leading making CPU’s only useful in few areas. Where a GTX 1060/RX 480 outperform these CPU’s in most renders. I’m more keen on getting a board with 6+ PCIe slots for GPU’s.
Still on the CPU side, Naples should be an interesting product from AMD, or would the APU be? either way i’m happy that AMD is back in the game and giving us choice, and Intel some competition. In the end no matter what CPU camp you are in (Intel/AMD) we win
I was batting around the idea of getting a vendor-overclocked Kaby Lake machine so as to have good single-core performance while closing the gap with Ryzen a little in rendering, so much for that (not a good move by Intel when the data is showing their first noticeable marketshare decline in years). I was talking about how it seemed to be the safe choice, but I don’t want to be out 1000+ dollars if the chip overheats and the new machine goes dead.
As per recent articles, it’s been said that we’re not looking at standard X86 fare with Ryzen, and it’s possible that there’s still quite a bit of a performance boost to be had as we continue to see new BIOS updates (so if you’re watching on the sidelines, it might be a smart move). Ryzen may actually become that no-brainer upgrade choice after all.
First look at AMD’s “Whitehaven” CPU’s (up to 16 cores).
With a boost clock of up to 3.6 gigs and quad channel RAM support, it looks like it will be an absolute beast in rendering (but obviously the trade-off is slower single-core speed). These chips will go against Intel’s highest-end offerings in the i7 line.
More information, we know the 16 core chip itself will be named ‘threadripper’
What is really impressive is that this is all one chip (it’s not a dual chip setup at all), the combination of that and the higher number of lanes is set to make it the fastest general PC chip ever created,
Depending on price, which one would be the better choice, Vega card with 16gb memory or this CPU with 32 GB ddr4 memory. On one hand you only need a new GPU to max out render times, CPU is irrelevant in that case. On the other hand having a beast CPU to render with upgradable RAM for future render consumptions.
Threadripper? Is that confirmed? I thought the name “Ryzen” was juvenile, but that’s yet another level. Who makes these decisions? It must’ve been the guy that designed this case:
AMD seems to be really going after Intel in the HEDT market. Threadripper has more cores at about the same clockspeed, more PCI lanes at the lower end (44 across the board versus 28 on the lower end of Intel’s HEDT chips). If the prices are right they could absolutely annihilate Intel in this space. Where clock speed isn’t as important as core count. The only real failing is that AMD’s chips don’t overlclock very well. My Haswell-E chip goes to 4.5 with an AIO at low voltages. However at 4.0-4.1 but with more cores I’ll take the compromise. It’s getting harder to overclock the E chips anyway. The Broadwell-E chips didn’t have the same headroom as the Haswell-E chips in-terms of overclocking for me. I could only get up to 4.2 (could have been the chip I tested though. Silicon Lottery, etc).
Either way Intel is going to have some real competition in the HEDT space and that’s a good thing.
Threadripper is apparently the official name for the HEDT Ryzen’s. Along with Epyc for the server chips. Someone in AMD’s marketing team is having way too much fun.
It’s weird to see such a “serious” company have such a wacky product names but I really dig the threadripper.
If only they named the ryzen cpus phoenix
And I think it’s a smart move. Yes, it’s silly but at least ppl are talking about it. I didn’t see ppl talk about coffee lake or whatever that thing is called (as in the name itself).
I think it’s good that AMD’s engineers are having fun, it means they are fully engaged in their zeal to beat both Nvidia and Intel (the Threadripper chips have as many threads as the majority of Intel’s Xeon models, but with the clockspeed and design of a general PC chip).
