Will AMDs upcoming 16 core CPU be a big step up for CG rendering?

>>566197
Oh yeah,actually cant wait to get my hands on one especially for the price.

>>566197
Isn't Vega more exciting for 3D? The HBCC claims to create a virtual memory space that's hundreds of terabytes large. Wouldn't that allow for rendering or simulating more complex scenes on a consumer GPU

What do you think they will price the 32 core version at? 16 core bulldozer opteron was about $1600.

>>566197
Might be, GPU rendering is in some ways ahead, but getting this many more cores for such a low price seems very appealing.

>>566201
From what I understand, the optimizations for Vega revolve around enhanced caching of data to keep the cores fed, but how effective this will translate to rendering depends on whether it not caching is a bottleneck. Seeing as renderers seem to receive linear improvements with core/card count, I would say it's fast enough. Unlike a game, a renderer already knows ahead of time what it needs to do, while games need to react to the player in real time, the issue is simply crunching that data.

>>566289
It's basically two 1700's stitched together, so it might be priced anywhere from $700-1000. Remember that we may see multiple binned variants at different price points that are all basically the same chip.
Unlike Intel, AMD's architecture is scalable as far as you want to take it, so it's not like Xeons that require special engineering compared to the desktop chips. As long as they have a good yield, they may be able to churn these out quite cheaply.

What's really exciting is being able to put two of these in a system, having 32c 64t for possibly under $2000 is an absolute killer, especially since we know that with sufficient cooling, Ryzen cores can maintain turbo speeds across all cores.

Most software can't even take advantage of 8 cores. What is the purpose of a 16 core CPU one?

>>566295
>Most software
Have you ever rendered/simulated anything at all?

>>566298
Yes

Most simulation software don't scale at all.

>>566299
Then you have NEVER used any simulation software. What a n00b! LOL! Imagine simulation software uses only 1 core on your PC. Nothing would get done even if it was networkable. You would have to own a few thousand PC's all networked together to get shit done.

>>566299
If by most you mean Houdini, you are wrong.

>>566294
>What's really exciting is being able to put two of these in a system, having 32c 64t for possibly under $2000 is an absolute killer, especially since we know that with sufficient cooling, Ryzen cores can maintain turbo speeds across all cores.
You don't need two, they are selling 32 core with 8 channel memory and 128 pcie lanes in a single e-atx board. The only question is if they sell them at a reasonable price for 4 dies + interposer or if they just go slightly lower than intels stupid 4k per cpu pricing.

Wouldn't Intel's i9 not the better option with the way superior single core performance after all?

I hope this does extremely well so intel would come down from their fucking high horse and lower the price... Intel cpus are unthinkably pricey... especially if you live in 3rd world...

>>566312
Difficult to say, in the past intel did not cut prices when AMD offered faster CPUs for lower prices.

>>566294
>with sufficient cooling
>AMD hardware
Maybe on another planet.

>>566299
Houdini does, and that's the top dog for simulation (probably in big part precisely because it scales so well).

>>566299
Wut. Maybe that's true for Blender, but even they are using OpenMP for a bit of multithreading.

>>566371
Ryzen runs significantly cooler than Intel. They solder the heatspreader, which is close to the best solution. The best being some strange metal TIM that nobody's heard of. Intel uses shitty consumer-grade TIM, which is why delidding and relidding is a thing.

>>566452
>Ryzen runs significantly cooler than Intel
Maybe on another planet. But I agree AMD runs cooler than the sun.

>>566452
>Ryzen runs significantly cooler than Intel
three AMDollars have been deposited into your account

>>566197
1. No 3d program used in the entertianment field today is written for a multi-core system. There isn't even 3d suite that takes advantage of gpu cores. Unity and Unreal have no plans to even develop for this.

2. AMD has done nothing to push Vulkan, OpenCL or any other programing endeavors that would advance core usuage. AMD doesn't even do it's own R&D anymore!

3. AMD and ATI fragmentation has made quality products a crapshoot. No one knows anymore what hardware is best suited for a job, it's always the hot new thing that has features nobody is asking for

4. The render-farm market has tanked. Most app developers have not made a major update solely because of 1. Local servers are, surprisingly, getting more expensive to build and maintain because of the cloud fad.

5. Homosexuals and millenials have ruined open source development for core usuage. Idenditity politics have become the de facto marketing stradegy for getting the word out, thus poisioning any real discussion

File: baitgis.jpg (40KB, 960x737px) Image search: [Google]

40KB, 960x737px

>>566500

>>566503
it already has

File: 1496171004749.png (871KB, 801x1500px) Image search: [Google]

871KB, 801x1500px

>>566464
>>566486
http://www.tomshardware.com/news/intel-x-series-skylake-x-kaby-lake-x-x299-basin-falls-core-i9,34545.html

>The LGA2066 socket is compatible with LGA2011 cooling solutions, which is a plus. However, Intel now recommends water cooling as the minimum cooling option, which will add more cost to the overall solution.

>However, Intel now recommends water cooling as the minimum cooling option,
>minimum cooling option
>minimum
get with the times gramps

>not getting into the i9 train

I'm really looking forward to a 2x16 core Epyc build.
>but anon, why not a single 32 core build?
Hopefully higher clocks on some of the 16 core SKUs

>>566510
teleports behind you
unzips katana
nothin personel kid

>>567850
But that's the cool thing about the Ryzen architecture, all dies are created equal and only binning determines their voltage efficiency, and by extension, the max clocks.
Assuming AMD doesn't pull that Xeon shit and lock down their CPUs aside from the BCLK, you should be able to run everything up to the 32-core model at high all-core turbo speeds as long as the cooling can handle it.
Considering the 1700x runs at 4Ghz on all cores with any decent air cooler, I think a D15-sized cooler should be able to hold down 16 cores quite well, and 32 cores on water loop. Just imagine a dual-socket 64-core system running at 4Ghz, it just might be possible.

>>566917
i9 chips are horrible.

File: IMG_5819.jpg (169KB, 1009x1055px) Image search: [Google]

169KB, 1009x1055px

Holy shit...

sauce pls

I can also only imagine what a field day I'm gonna have in eBay buying up used xeons when data centers start upgrading in a year

>>567864
sweet jesus you're right...

>>567864
>AMD
>as long as the cooling can handle it

File: IMG_5824.jpg (47KB, 614x360px) Image search: [Google]

47KB, 614x360px

>>568372
https://www.bit-tech.net/hardware/2017/06/16/intel-core-i9-7900x-and-x299-chipset-revie/8
>However, temperatures were definitely a concern with Cinebench and Terragen pushing 100°C with our 240mm AIO liquid cooler.
Would you prefer a 10-core Intel housefire?
Don't forget the following facts:
TR CPU is significantly larger in area = better heat dissipation
The IHS is soldered to the die, not pudding like Skymeme-X
Much lower power consumption

The proof will be in the benchmarks, but Ryzen has a lot going for it right now.

>>568398
Don't forget the following fact: you're either an advertiser or you have some really weird hardware-based fetishes. And AMD is still garbage.

>>568438
>you have some really weird hardware-based fetishes
Like what, supporting or shitting on companies based on the quality of the products they make, rather than fanboyism? Oh yeah, I forgot this was 4chan for a moment there.

Don't get me wrong, I agree that AMD used to be garbage for a pretty long while now, but it's also silly to discount what they have now simply based on their recent history. Look at banchmarks, not labels. If I can get a processor that's as fast as the other guy's but is half as expensive and also runs cooler, I'll even accept having AMD written on the box.

>>568438
>AMD is still garbage
Go and tell this to /g/, they will slaughter you alive.

>>568456
I just did. I don't know how to link to it. I'm going to try: >>/g/60948674 Let's see how it goes.

>>568451
I'm with this dude..
why people lost their shit when they saw raizen? it's like they really want intel to be worse than fucking AMD... because, muh corporations...
I swear feminism influences every weeb on earth... (I'm not even against it DESU)

>>568476
what do feminism and weebs have to do with CPUs?

File: hqdefault.jpg (15KB, 480x360px) Image search: [Google]

15KB, 480x360px

>>568492
Intel financially backs Anita Sarkeesian and feminist frequency.

>>568476
yeah why would we prefer AMD being a decent competitor instead of Intel having a monopoly?

File: IMG_5837.jpg (1MB, 2906x2299px) Image search: [Google]

1MB, 2906x2299px

So Epyc dropped recently and here are the prices. The 32-core single socket chip will be $2100. All chips feature all expansion capabilities; 8-channel memory for up to 2TB RAM per CPU, 128 PCI-E lanes, and also 64MB of L3 cache. Performance of one 7551P is claimed to be 20% above a dual E5 2650 v4 setup

>>568908
"Epyc" sounds like some 90's kid's IRC nick.

>>568908
There doesn't appear to be any worstation boards announced. The single 32 core might be interesting but if they don't let users change voltage and clocks it may only be 50% faster than threadripper.

>>568908
>16 core at 650 dollars

It's over. Shitntel is finished.

>>568551
if this is true it seems like a bad omen. for intel that is.