redtuesday

There are even rumors that Intel delayed the much more important Xeons to 2020, which could give AMD a 1 year head start with Zen2 on 7nm in the server space (comparable to Intels 10 nm).

Copy paste from my posting yesterday:

EDIT: it seems the twitter post got deleted.

EDIT2: anandtech still has the pictures: https://www.anandtech.com/show/13119/intels-xeon-scalable-ro...

If the rumor [0] that Intel's first 10nm server chip will only release mid 2020 is true, then AMD's shares will probably skyrocket again if they truly can release their Zen 2 server CPU mid 2019 (on 7nm which is comparable to Intel's 10nm).

[0] https://twitter.com/david_schor/status/1022142835989118977

If you want that I recommend to look at semiwiki [0] and wikichip [1][2]

[0] https://www.semiwiki.com/forum/content/7343-leading-edge-log...

[1] https://en.wikichip.org/wiki/10_nm_lithography_process

[2] https://en.wikichip.org/wiki/7_nm_lithography_process

The announcement: https://steamcommunity.com/app/282590/discussions/1/17106901...

He also has a three videos series about it on youtube.

https://www.youtube.com/user/DrBartosz/playlists

I (and many others) would take a 33% increase in core count over a 10-15% advantage in single core workloads. Even the gaming comparison is most of the time overemphasized. Does it really matter if you are getting 88FPS or 80FPS if your monitor refresh rate is capped at 60FPS?

Luckily it seems AMD can probably deliver both with Zen 2. For Epyc the roadmap suggest a 50% increase in core count [0] over Zen 1, and since Ryzen and Threadripper uses the same die as Epyc this will most likely trickle down to Ryzen 3xxx and Threadripper 3xxx.

Depends on your game and monitor I guess. Quite a few people seem to buy ugly 144hz gaming monitors and some games like Warhammer and Far Cry 5 can have ~20 fps difference at the same clocks at 1080p.

I feel that most of these attacks at AMD and their CPU-s tend to bend the truth a lot, and that isn't surprising, given the shady tactics that Intel has employed previously.

Most of the time the limiting factor is not the CPU but the GPU anyway. The good thing about AMD closing the gap or potentially even overtaking Intel in single threaded performance is that the fanboys, fangirls and corporate shills have one argument less for spreading FUD and suggesting Intel over AMD.

[0] there are even (unlikely) rumors floating around about a 64 core Epyc, but I don't see that happening before 7nm+ on EUV.

Do you mean the single threaded part or that OP claims that it's what most users desire? If it's about the single threaded performance then OPs claim is true.

default clocks: https://i.imgur.com/dU3aI35.png

normalized clocks: https://i.imgur.com/wg3SQUd.png (sadly they have not normalized memory clocks)

In single threaded games this can be even more apparent. But Intel is mainly faster because of a significant clock advantage. Clock for clock the advantage is small.

I'm looking forward to Zen 2 which could allow AMD to close the IPC gap or even get a bit ahead of Intel. Combined with the 7nm process which should allow better clocks it's looking good for AMD.

Question is if AMD uses the high performance process which is touted to allow up to 5 ghz [0] at the cost of increased power usage, or if they use the low power process (which is also more affordable). Since Ryzen, Threadripper and Epyc use the same die probably the latter since power consumption is so important in the data center.

[0] https://www.anandtech.com/show/12438/the-future-of-silicon-a...

I don't really follow the ML space, but when you have a good GPU in your PC (which I assume quite a few people here have) you can learn/use Tensorflow by running things locally instead of paying for Google Cloud etc. Why pay extra money if you already have a good GPU in your computer (unless you really need to decrease the time it takes to run the computation etc.)

If the rumor [0] that Intel's first 10nm server chip will only release mid 2020 is true, then AMD's shares will probably skyrocket again if they truly can release their Zen 2 server CPU mid 2019 (on 7nm which is comparable to Intel's 10nm).

[0] https://twitter.com/david_schor/status/1022142835989118977

AMDs Zen architecture uses 32KB for data and 64KB for instructions (was curious if there are differences between AMD and Intel designs regarding L1 cache).

Since I read about this a few years ago I really want to try it, but I don't want to buy an extra GPU for it. I hope AMD [0] brings their SR-IOV implementation called MxGPU down to their mainstream GPUs, which allows to split a single GPU between host and guests. Apparently this would also be more secure than passthrough?

In order to not affect their pro GPU sales they could maybe limit the number of virtual GPUs from 16 to 2, which would be enough for the host and a single guest.

[0] or Nvidia, I don't care, but since Nvidia is the market leader they have less incentive than AMD.

How would you prevent people who don't even have that from buying these bikes?

I never took the time to investigate how accurate this site is, but it seems quite nice since it has uv index, time to skin redness and time to sunburn:

http://sunburnmap.com

[...] even kids are sent to play covered in full sleeve clothing.

Which is good because if I remember correctly the human skin has quit a bit natural sun protection, but is only fully developed in the twenties.

If I got that mixed up with something else please correct me if I'm wrong.

Can't you just combat fork bombs with e.g

  docker run --pids-limit=64

As a layman: why don't things like GridCoin take off? Is it only mindshare/being late to the party/or something alike, or are there technical reasons for it?

Bitcoin seems like a waste of energy in comparison (again, as a layman). Apparently enough to be in the top 65 of global power consumption for countries. [0][1]

[0] https://digiconomist.net/bitcoin-energy-consumption

[1] https://en.wikipedia.org/wiki/List_of_countries_by_electrici...

To use the plus sign like behaviour you now need to use quotes (since 2011). Instead of:

  mars +landings +failures

  mars “landings” “failures”

It seems to be related to php. A user on reddit which has a Epyc 7551 CPU has no segfaults without the php portion of the test: https://www.reddit.com/r/Amd/comments/6rmq6q/epyc_7551_minin...

To add to that:

http://www.hardwarecanucks.com/forum/hardware-canucks-review...

https://www.hardocp.com/article/2017/08/10/amd_ryzen_threadr...

http://www.gamersnexus.net/hwreviews/3015-amd-threadripper-1...

http://www.legitreviews.com/amd-ryzen-threadripper-1950x-thr...

http://www.techradar.com/reviews/amd-ryzen-threadripper-1950...

http://hexus.net/tech/reviews/cpu/108628-amd-ryzen-threadrip...

https://www.vortez.net/articles_pages/amd_ryzen_threadripper...

google translate of computerbase.de: https://translate.google.com/translate?hl=de&sl=de&tl=en&u=h...

google translate of pgh.de: https://translate.google.com/translate?sl=de&tl=en&js=y&prev...

google translate of tomshardware.de: https://translate.google.com/translate?sl=de&tl=en&js=y&prev...

google translate of hardwareluxx.de: https://translate.google.com/translate?sl=de&tl=en&js=y&prev...

Like I have written in the other thread already: add to that the ECC support of Threadripper [0], 20 more PCIe lanes and no raid key shenanigans (on Intels x299 you have to pay 100$ extra for RAID 1 and 300$ for RAID 5 support with their VROC feature [1]). Unless you really need AVX 512 and better single thread performance (and 2 more cores in case of the not available 18 core part), why would you buy Intels new offering? Because you can reuse the cooler since it's compatible with x99?

[0] https://www.reddit.com/r/Amd/comments/6icdyo/amd_threadrippe...

[1] https://youtu.be/TWFzWRoVNnE?t=11m38s

Which Intel part has 36 cores? Do you mean threads, or is there a Knights Landing card with so few cores I'm not aware of?

Threadripper 1950X with ECC, 16 instead of 10 cores, 64 instead of 44 PCIe lanes and without a raid dongle which prevents certain raid modes unless you pay for them (like with Intel): $999.

I love competition. Though I would prefer if there was a third competitior in the x86 CPU space and the GPU space.

Threadripper will have ECC support [0] just like Ryzen but validation depends on the motherboard manufacturer.

[0] https://www.reddit.com/r/Amd/comments/6icdyo/amd_threadrippe...

Unfortunately without delidding you're not going to hit good OC clocks given the current TIM. And delidding is a dealbreaker on a $1000 CPU.

Yeah, absolutely agree. What's sad is there are fanboys defending this and saying this makes direct die cooling easier through delidding, which saves 1-2 C° over solder, and is the beast idea Intel had in recent years.

Reading the text I think they talked about the chips with a core count larger than 10, which were never on any roadmap until AMD released Ryzen (12 core Skylake appeared on the Roadmap) and then announced Threadripper (14, 16 and 18 core Skylake X paper launched).

Not that I know of, but at their analyst day in May 2017 AMD showed slides in which they beat Nvidias GP100 in Baidus Deep Bench with their new GPU Vega. [0][1] Since AMD recently released the Vega Frontier Edition [3][4] I posted this here in the hopes to see some benchmarks from users here on HN.

[0] http://i.imgur.com/twhTpcC.jpg

[1] http://i.imgur.com/1peXVnq.png

[3] tldr: 16 GByte HBM2, 25 TFlop FP16, ~1000-1500 dollar.

[4] https://www.newegg.com/Product/Product.aspx?Item=N82E1681410...

Looks like the title was changed. Yesterday it was called "AMD deep learning library ROCm 1.6 released (Caffee support and more)"

It seems at the moment only Caffe is public and it's a port of the CUDA version to OpenCL [0][1] to allow it to run on AMD GPUs like the recently released AMD Vega Frontier Edition. From the wording on the page it looks like it could be an equivalent to CuDNN.

[0] through AMDs tool called HIP which can convert CUDA

[1] http://gpgpu10.athoura.com/ROCM_GPGPU_Keynote.pdf

I like the concept. At the moment I use contentful so non technical people can edit their static site and don't have to worry about running a server (and especially keeping the software running on it up to date), but I'll probably switch to NetlifyCms (or hopefully emerging competitors) when they mature a bit.

The thing I like the least at the moment is form submissions.

As a layman which also thinks of mining as a waste of engery (although it seems necessary as the other commenter pointed out) I hope gridcoin [0] or something similiar takes off.

[0] http://www.gridcoin.us/

Someone archived the page, so for the people getting a 403: http://archive.is/362lF

Yes, but what I meant is uneven CCX. The 4 core parts have 2+2, the 6 core parts have 3+3, the 8 core parts have 4+4. There is no uneven CCX combination like 2+4 for example.

If this applies to multi die solutions like Threadripper, there will be no 10 and 14 core parts as they would require uneven CCX combinations.

If the Epyc lineup is an indication this might be true. Especially since there is no 12 core part for Epyc. Since it has 4 dies you could produce 12 cores by using only 3 cores of each die, but this would require uneven CCX (1+2 or 0+3).