AMD processors?

Message boards : Number crunching : AMD processors?
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Phud Redux

Send message
Joined: 20 Apr 16
Posts: 270
Credit: 2,976,272
RAC: 1
United States
Message 1836134 - Posted: 15 Dec 2016, 21:07:44 UTC

Are they really at such a disadvantage compared to Intel CPU's?
ID: 1836134 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1836138 - Posted: 15 Dec 2016, 21:39:33 UTC - in response to Message 1836134.  

You might conclude that opinion if all you did was CPU tasks or were concerned about wall power usage. Perfectly capable of feeding GPU tasks to your graphics cards which do most of the heavy lifting for a great many projects.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1836138 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22200
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1836144 - Posted: 15 Dec 2016, 21:51:00 UTC

Generally AMD (FX-series) processors are OK for feeding GPUs, but are about two-thirds the performance of an i7
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1836144 · Report as offensive
Profile RueiKe Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 14 Feb 16
Posts: 492
Credit: 378,512,430
RAC: 785
Taiwan
Message 1836188 - Posted: 16 Dec 2016, 0:39:57 UTC

They can definitely perform well in feeding significant GPU compute capability. I have two FX-8370 based systems that are in the top 5 computers ranked by SETI RAC.
GitHub: Ricks-Lab
Instagram: ricks_labs
ID: 1836188 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1836230 - Posted: 16 Dec 2016, 6:11:04 UTC - in response to Message 1836134.  

Are they really at such a disadvantage compared to Intel CPU's?

For work done at a given clock speed, Intel CPUs leave them way behind. For power usage for a given workload, Intel CPUs leave them way behind. For work done & power used, AMD isn't even in the race.
However AMD are shortly releasing a new CPU architecture, and if any of the initial indications are even halfway accurate, AMD will finally be able to put some pressure on Intel again.
Grant
Darwin NT
ID: 1836230 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1836278 - Posted: 16 Dec 2016, 13:55:35 UTC - in response to Message 1836230.  

Are they really at such a disadvantage compared to Intel CPU's?

For work done at a given clock speed, Intel CPUs leave them way behind. For power usage for a given workload, Intel CPUs leave them way behind. For work done & power used, AMD isn't even in the race.
However AMD are shortly releasing a new CPU architecture, and if any of the initial indications are even halfway accurate, AMD will finally be able to put some pressure on Intel again.

It's been about a decade and a half. So they are due to release something competitive again . I didn't mind the days of getting top end Intel CPUs for ~$300 vs ~$1000 where they are back up to again.
Maybe they will even have ALUs with good performance this time. I'm expecting they will not, but we will have to wait to find out.

One thing AMD does have going for them is their APUs have better integrated graphics compared to Intel, but that's what buying ATI will do for you.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1836278 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34258
Credit: 79,922,639
RAC: 80
Germany
Message 1836297 - Posted: 16 Dec 2016, 15:58:05 UTC
Last modified: 16 Dec 2016, 16:01:10 UTC

I have watched the AMD presentation of RyZen a few days ago.
The Top model will be a 3.4 GHZ 8 core 16 thread CPU.
Looks promising.

https://youtu.be/4DEfj2MRLtA


With each crime and every kindness we birth our future.
ID: 1836297 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1836350 - Posted: 16 Dec 2016, 20:01:48 UTC - in response to Message 1836278.  

Maybe they will even have ALUs with good performance this time. I'm expecting they will not, but we will have to wait to find out.

As Mike mentioned, the early signs are very promising.
It doesn't look like a return the K7 glory days, but at the very least it appears they will be able to match current Intel performance & power consumption. That will allow AMD to charge more for their hardware, and improve their resources for R&D. And as you mentioned if they can put some pressure on Intel we should see them drop their process again to try and force AMD out of the market.

To me the most promising thing is they're not going over top with pushing the new architecture. Even since the Athlon CPUs first came out each new or tweaked CPU has been hyped for it's performance. And every time it's been either a disappointment, or a large disappointment (Bulldozer, Piledriver anyone?)
This time around it like they're going to let the hardware speak for itself, no need for over the top marketing hype and claims that are never met.
Grant
Darwin NT
ID: 1836350 · Report as offensive
Cruncher-American Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor

Send message
Joined: 25 Mar 02
Posts: 1513
Credit: 370,893,186
RAC: 340
United States
Message 1836351 - Posted: 16 Dec 2016, 20:04:23 UTC - in response to Message 1836134.  

Are they really at such a disadvantage compared to Intel CPU's?


Yes - some AMD processors nominally have n cores, but only n/2 floating point units, so (for SETI purposes) they cannot really process more than n/2 WUs efficiently. I tried an 8 core (FX) cpu on a test machine and found that it had real problems running > 4 WUs.
ID: 1836351 · Report as offensive
Phud Redux

Send message
Joined: 20 Apr 16
Posts: 270
Credit: 2,976,272
RAC: 1
United States
Message 1836357 - Posted: 16 Dec 2016, 21:10:13 UTC

i watched the video.
1. they talk about matching the i7-6900.
2. they talk about the i7-6900 price point of $1,100.
3. Intel is obviously working on the i-7 7000 and 8000 series as that film was shown.
4. even if the sell the Ryzen @ $500., Intel will drop the next shoe and probably for less.
5. a long uphill road for AMD, i think.
ID: 1836357 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1836358 - Posted: 16 Dec 2016, 21:24:36 UTC - in response to Message 1836357.  

5. a long uphill road for AMD, i think.

No doubt.
But it will be the first time in over a decade that AMD has been competitive. And given that this is the first production release of a new architecture, there will be room for improvement. Look at where Intel is now since the introduction of the original series of Core processors.
Grant
Darwin NT
ID: 1836358 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1836390 - Posted: 17 Dec 2016, 0:42:21 UTC

I don't remember where it was that I read this, but there was a UK tech site that said the experts and veterans of the tech reporting industry seemed to be in the mindset that the top level Zens (8c/16t) would likely be in the £275 range. So it seems like the 8c/16t Zen is going to be within +/-10% the performance of 6900k at a fraction of the price. Even if Intel drops the 6900k price in half, AMD is still in a reasonable position to be ~$200 cheaper for nearly identical performance.

Plus, AMD's hail mary they're going for is not the desktop market, but instead the server market. Naples looks good, and they have good plans for the near future that can be a lot better than Xeons.

Like this gem: http://wccftech.com/starship-amd-processor-48-cores-7nm/

And wccf tech seems to agree with what I read a few months ago about pricing: http://wccftech.com/amd-zen-summit-ridge-prices-details-leaked/
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1836390 · Report as offensive
AMDave
Volunteer tester

Send message
Joined: 9 Mar 01
Posts: 234
Credit: 11,671,730
RAC: 0
United States
Message 1836391 - Posted: 17 Dec 2016, 0:52:17 UTC - in response to Message 1836278.  
Last modified: 17 Dec 2016, 0:52:40 UTC

One thing AMD does have going for them is their APUs have better integrated graphics compared to Intel, but that's what buying ATI will do for you.

Message 1835829
See last bullet
ID: 1836391 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34258
Credit: 79,922,639
RAC: 80
Germany
Message 1836465 - Posted: 17 Dec 2016, 9:13:27 UTC - in response to Message 1836390.  

I don't remember where it was that I read this, but there was a UK tech site that said the experts and veterans of the tech reporting industry seemed to be in the mindset that the top level Zens (8c/16t) would likely be in the £275 range. So it seems like the 8c/16t Zen is going to be within +/-10% the performance of 6900k at a fraction of the price. Even if Intel drops the 6900k price in half, AMD is still in a reasonable position to be ~$200 cheaper for nearly identical performance.

Plus, AMD's hail mary they're going for is not the desktop market, but instead the server market. Naples looks good, and they have good plans for the near future that can be a lot better than Xeons.

Like this gem: http://wccftech.com/starship-amd-processor-48-cores-7nm/

And wccf tech seems to agree with what I read a few months ago about pricing: http://wccftech.com/amd-zen-summit-ridge-prices-details-leaked/


The CEO from AMD clearly said that the 8c/16t CPU is for the gamers market on AM4 plattforms thats not the server market.


With each crime and every kindness we birth our future.
ID: 1836465 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1836544 - Posted: 17 Dec 2016, 19:02:57 UTC - in response to Message 1836465.  
Last modified: 17 Dec 2016, 19:16:47 UTC

Plus, AMD's hail mary they're going for is not the desktop market, but instead the server market. Naples looks good, and they have good plans for the near future that can be a lot better than Xeons.

Like this gem: http://wccftech.com/starship-amd-processor-48-cores-7nm/

The CEO from AMD clearly said that the 8c/16t CPU is for the gamers market on AM4 plattforms thats not the server market.

Yes, that is certainly correct. But the Zen architecture is also the same design that is going to be used on Naples in the server line, but on a different socket, of course. I read press articles from about this time last year and AMD is in so much debt at this point that the enthusiast market isn't really going to make a dent, but if they can pry 10-15% of the server market away from Intel with a competitive product, that can literally save AMD from having to sell off 75% of the company's assets. They're over $19bn in debt and have made deals to defer payments until 2019, at which point, the debt is owed 100%.

Yes, Summit Ridge is for gamers basically, but Naples (the same CPU, but with a server socket) is intended to be the real hero for AMD's future.

Also interesting to note.. if I read the specs and a few articles correctly, the new Xbox One S has a 16c/32t Zen in it, which is a SOC Naples CPU. Back in the summer, it was said that the One S was going to have ~10TFLOPs of total compute power, and we know how much the RX 480 does--5.8, which means the CPU should do 4.2. At that time, I thought it was an 8c/16t Zen, which meant it was going to be ~250GFLOPs per core. Now that I know there are twice as many cores, that puts it at 125GFLOPs/core, which is still way above the 85 you get out of an FX-8350 @ 4.8GHz, but a bit behind ~160 for a high-end i7 (I didn't look for a specific model, I just googled 'gflops i7' and read the result page looking for a GFLOP figure). But again, it's a Naples server CPU, so it probably isn't running at 4+ GHz and is more likely down in the mid 2's.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1836544 · Report as offensive
Al Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 3 Apr 99
Posts: 1682
Credit: 477,343,364
RAC: 482
United States
Message 1836582 - Posted: 17 Dec 2016, 21:41:05 UTC - in response to Message 1836544.  

They're over $19bn in debt and have made deals to defer payments until 2019, at which point, the debt is owed 100%.
:-O

I had no idea. If you were top level management there, how could you sleep at night? Talk about sink or swim. Intel pulls the next gen out, or even starts seeding the market with some FUD or casual semi-vaporwareish talk, and that would probably be the end of AMD. Ouch. Though I don't think Intel is known for that behavior, unlike some other companies I won't mention.

ID: 1836582 · Report as offensive
Phud Redux

Send message
Joined: 20 Apr 16
Posts: 270
Credit: 2,976,272
RAC: 1
United States
Message 1836611 - Posted: 17 Dec 2016, 23:14:08 UTC - in response to Message 1836582.  

They're over $19bn in debt and have made deals to defer payments until 2019, at which point, the debt is owed 100%.
:-O

I had no idea. If you were top level management there, how could you sleep at night? Talk about sink or swim. Intel pulls the next gen out, or even starts seeding the market with some FUD or casual semi-vaporwareish talk, and that would probably be the end of AMD. Ouch. Though I don't think Intel is known for that behavior, unlike some other companies I won't mention.



intel needs competition. they will not admit it...but all r&d companies need competition. if nothing else to keep from being complacent.
ID: 1836611 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1836619 - Posted: 18 Dec 2016, 0:11:46 UTC
Last modified: 18 Dec 2016, 0:26:16 UTC

It is not in Intel's interests for AMD to fold or go under (AMD as a company won't, though--the GPU division is doing pretty well, and the contracts with Nintendo and Xbox are helpful for revenue, as well--so AMD won't completely go away). The main reason for this would be.. if AMD has to pull out of the CPU market... then Intel will then be in violation of global antitrust (monopoly) laws. Intel needs a competitor, and one that has a reasonable chance of taking a decent market share in order to abide by the laws and regulations.

To make matters worse.. AMD is paying royalties to Intel for x86 capabilities... and Intel is paying AMD royalties for 64-bit capabilities. They both need each other.

That said.. IF AMD had to pull out of the CPU market, they could sell their fab plants to another company that could be in a reasonable position to enter into the market as a competitor, and that would most likely be Samsung. Samsung hasn't expressed any interest in getting into the x86 processor market, but they are probably the most-capable of doing so.

I don't think it will come anywhere close to that though. AMD may not hit a grand slam out of the park (baseball reference) on this launch, but it seems like it could still be a solo home run, simply because if Zen is within ~10% of the performance of the 6900k at 1/3 of the price.. that's a no-brainer, you go for the cheaper one. Same for the server market. If you can get a 2.5ghz 16c/32t "Opteron" (I don't know if they're still using that name for the server line anymore) for.. even 75% (25% cheaper) the price of a 2.7GHz 16c/32t Xeon, and the Opteron is within about 10-15% the performance and you're about to place an order for 10,000 units... it's obvious which one you're going with--the cheaper one.

That's what AMD needs on this one. They don't need to one-up/leap-frog Intel, they just need to be tied or on-par with them for a fraction of the price, and it looks like that's exactly the position they're in.


edit: Okay, it was bugging me and I had to dig a little to find what I had read a while back. It's not nineteen (19) billion in debt, but it is still a lot of debt they're in, and payments are coming due in 2019. http://www.fool.com/investing/general/2015/08/25/zen-is-advanced-micro-devices-incs-last-chance.aspx
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1836619 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13736
Credit: 208,696,464
RAC: 304
Australia
Message 1836623 - Posted: 18 Dec 2016, 0:42:37 UTC - in response to Message 1836619.  

That said.. IF AMD had to pull out of the CPU market, they could sell their fab plants to another company

AMD are fabless and have been since around 2009, that's why they've been using TSMC for their CPUs & GPUs.
Oct 7, 2008. With AMD dumping its fabs, Intel (NASDAQ: INTC) and some of the Asian memory makers are the only chip companies with their own fabrication facilities. Everyone else outsources their semiconductor manufacturing.

you're about to place an order for 10,000 units... it's obvious which one you're going with--the cheaper one.

When it comes to the server business, that often isn't the case. Initial costs are a part of the considerations, but another significant consideration is power use.
What good is it buying a server with a low upfront cost, if over it's lifetime it costs 2, 3 or 4 times as much to run? And if the performance of a given server is only half that of another, you need twice as many to get the same performance; so the upfront cost needs to be at least half that of the better performing system, and that's without even taking in to account running costs.
Grant
Darwin NT
ID: 1836623 · Report as offensive
AMDave
Volunteer tester

Send message
Joined: 9 Mar 01
Posts: 234
Credit: 11,671,730
RAC: 0
United States
Message 1836632 - Posted: 18 Dec 2016, 1:17:31 UTC - in response to Message 1836623.  

That's what AMD needs on this one. They don't need to one-up/leap-frog Intel, they just need to be tied or on-par with them for a fraction of the price, and it looks like that's exactly the position they're in.

edit: Okay, it was bugging me and I had to dig a little to find what I had read a while back. It's not nineteen (19) billion in debt, but it is still a lot of debt they're in, and payments are coming due in 2019.

I believe you were substituting the year ‘19 for AMD's debt.  It is significant, but no where near $19B.  Scroll to the paragraph titled “Balance Sheet & Debt” in the following link:  AMD Is Cooked - Debt Analysis & Forecasting - Part 3

When it comes to the server business, that often isn't the case. Initial costs are a part of the considerations, but another significant consideration is power use.
What good is it buying a server with a low upfront cost, if over it's lifetime it costs 2, 3 or 4 times as much to run? And if the performance of a given server is only half that of another, you need twice as many to get the same performance; so the upfront cost needs to be at least half that of the better performing system, and that's without even taking in to account running costs.

Take a look here.
Excerpt:
    ♦  The 8 core, 16 thread RYZEN CPU has a TDP of 95 watts, 45 watts less than its direct Intel competitor.

    ♦  The 8 core 16 thread RYZEN CPU competes with Intel’s fastest 8 core 16 thread offering, the $1199 Broadwell-E. i7 6900K, in terms of features and performance and exceeds it in power efficiency.

ID: 1836632 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : AMD processors?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.