AMD Launches 64-core / 128 Threaded 2nd Gen AMD EPYC Processors
Click here to post a comment for AMD Launches 64-core / 128 Threaded 2nd Gen AMD EPYC Processors on our message forum
TieSKey
Oh, completely forgot about that. Makes sense.
waltc3
This is a 1-2 punch that Intel may never recover from, imo. They simply don't offer competitive products. The server market is where the real money is in the computer markets--big companies buy hundreds and even thousands at a time, need 24/7 reliability, Registered DIMMs, 8-channel memory interfaces, and gobs and gobs of PCIe lanes--and security is paramount--that's the other gigantic edge AMD has over Intel--security--it's not even close there, either! According to the presentation, AMD already has Zen 3 design-complete, and work on Zen 4 has begun. If Intel's waiting on AMD to drop the ball as it did post the A64/Opteron era, I'm afraid they are going to have a very long wait.. While Intel was busy milking old architectures and raking in the dough--AMD was busy designing Zen and Epyc--what a difference in corporate management! By the time Intel "has an answer" to Zen2, AMD will be at 5nm on Zen3 or Zen 4! Intel looks to be playing catch-up for quite some time to come.
waltc3
[QUOTE="schmidtbag, post: 5698489, member: 246171
....
That's definitely enough time for AMD to gain enough marketshare to fund an even better architecture.
What?....;) Have you forgotten that when AMD's marketshare was a competitive joke 4-5 years ago and they were living on canned beans and peas that the company had plenty of money to design Zen and Epyc regardless, and bring it to market? AMD's got everything it needs to go full bore on Zen 3, right now. People are buying--or, rather they should be buying-- AMD stock like there is no tomorrow, imo. Papermaster said at the presentation that work on Zen 3 is already design-complete, even as we speak. The one thing AMD does not need is "marketshare to fund their next cpu development"...;) They are already knee-deep into Zen3. Several AMD executives reiterated the fact that AMD is here for the long haul and has no intention of occupying 2nd place ever again--Papermaster even said it was "personal" for him, as did Norris. Just wait 'till to you see AMD's Q3 and Q4 quarter for 2019, btw. The second-quarter's results ended prior to 7/7. Should be real eye-poppers, I'll bet.
anticupidon
It's too late to buy AMD stocks?
schmidtbag
I don't think you understood what I was trying to say.
Right now, AMD is demolishing Intel in terms of performance and value in the server market. Just because choosing them is a no-brainer, doesn't mean all servers are suddenly going to switch to them. Despite what a lot of people think, AMD is probably not going to have a greater market share than Intel in the next 2 years. But... they will most likely take the majority of new server hardware sales (at least for x86 platforms).
All that being said, AMD has had sufficient funding to top Intel, but it's naive to think Intel isn't going to eventually catch up. However, when Intel inevitably makes a compelling competitive product, AMD will have had enough time and sales to maintain their competition.
TL;DR
I'm talking about what's going to happen 5-10 years from now. Today, AMD is the obvious choice for most servers, but they won't remain that way forever.
D3M1G0D
You're welcome to pile in - the stock jumped a massive 16.2% today due to Rome (and Google and Twitter's adoption of it). The more the merrier 😉
Andrew LB
One of the most important things about advertising.... if you have to mention your competitor, you've already lost.
Andrew LB
AMD has about 2% datacenter market share. up from 1.3% last year. I had to read through over a dozen articles to find that number instead of the nonstop misleading "AMD gains 56% in data center....". Doing this kind of deceptive reporting will result in an overvaluation of their stock and will eventually cause lots of people to lose money. It seems the tech press is just as divorced from reality as the political press these days.
Astyanax
except they're winning all the internets.
schmidtbag
Where are you getting 56% from? The article you linked to mentioned 1 source that only predicted they would take 25% marketshare, and was vague about timeline. Last time I checked, that's a lot lower than 56%.
But even then, there's a big difference between how much of the existing market is divided vs how many new systems are being sold. So, it's actually totally reasonable that AMD could be getting 56% marketshare in datacenter sales in the last quarter, half, or full year.
Also, these new generation of Epycs were just released. Of course they haven't affected marketshare numbers yet because there's no collected data on them yet.
Gomez Addams
I believe that phrase meant their market share improved 56%. For example, if they had 10% and grew to 15% they gained 50%.
Gomez Addams
This kind of thing will be of little use to most people but I can definitely make use of it. Currently I have to use networks of four or five computers to do my computations and this could potentially replace all of that 20U monstrosity. We tried one of Intel's boxes with 20 cores and 40 threads and it wasn't quite up to the task. I needed two of them because their clock rate was too slow (2.2GHz) and the app didn't scale far enough because of cache and memory thrashing. I need at least 30 cores going full blast to meet my timing requirements and I get that with five boxes (hence the 20U remark - 4U each) but it would be great to have it in a single 1 or 2U box. Here is a picture of the task manager on the 20 core machine.
https://i35.servimg.com/u/f35/17/98/38/10/taskma10.png
Hog54
https://markets.businessinsider.com/news/stocks/amd-stock-price-epyc2-chip-turning-point-for-silicon-valley-2019-8-1028431822
mohiuddin
I
I'll never support this kind of advertisement . It will just spread hate among companies and customers , no healthy competition
Aura89
I'm assuming he actually meant they stated AMD marketshare rose by 56%, which isn't misleading, i don't know why anyone would think that means they have 56% marketshare.
In the quote you posted, he stated the article shows 1.3% to 2% marketshare, which is just about 56% (it's actually 54%) of an increase.
Gomez Addams
I just realized something about this new chip. I mentioned previously that my app didn't scale well on the 20-core machine because of memory and cache thrashing. I looked up the specs of that chip and it has 28MB of cache. AMD's highest-end chip has 256MB of cache. That's amazing! It's about 9X more. I've watched the memory usage of my app and I think it could fit almost entirely in that cache. That would let it really scream a on machine with that chip. I think it would be very interesting to try one.
BTW - that machine with the 20 cores had two chips in it so that's why there are 80 boxes in the previous shot : 20 cores x 2 threads each x 2 chips.
asturur
Hopefully who buys stocks know what data to look at.
They trust press release when they say `this year we did shit`.
asturur
I'm not at all a software expert. But if you can't meet your expectation with 40 cores, that may be that the app ( the task the app has to solve or the implementation ) is not a fit for multi threading. The 256 megs of cache are to be divided by the 8 chips, so is like 32 every 8 core, not a super jump compared from the 28 of xeon.
Cerb3rus
lets see if dont have compatilibity issues like Ryzen =)
Richard Nutman
Some tasks just need huge amounts of processing power.
We make frame rate conversion software where I work, and for the quality and to be in real-time we use three top end NV cards all crunching away in openCL.
We looked at doing a pure software version, but didn't get close with a dual Broadwell Xeon machine.
Perhaps with 8 channel memory and dual Rome machine, it may be possible.
