The Ryzen "ThreadRipper"... 16 cores of awesome

[USER8000]

AMD needs some way to into that market. Yes, that is the key to their success. But price is not the way to compete there. Features, stability, reputation, quality, etc. That is how you get into that market. ECC memory is one massive feature to tout. They don't also need price.

What AMD needs is to have more money to get it into Dell's hands, validate the crap out of it, and bring those results to potential customers. That costs money that AMD doesn't have and will not have unless they start charging for it.

The phrase "no one has been fired for buying IBM" has been applied many times to Intel as well. If you are the one recommending, approving, or buying 20,000 computers, then you better believe that you will double think what will happen if something goes wrong and the CEO asks why you didn't buy the name brand. Remember, the CEO (or similar person of power) probably rates wine taste by the price that they were told that the wine cost (regardless of the actual quality of the wine). http://www.telegraph.co.uk/news/sho...-better-if-they-are-told-it-is-expensive.html, and for similar reasons thinks that Apple makes the best everything, Dyson makes the best vacuums, Starbucks has the best coffee, etc.

On the low end, low price is killing AMD. For decades, AMD had the cheap chips and OEMs put them into cheap computers with cheap motherboards and cheap components. Guess who gets blamed if something is slow, bad, or malfunctions? AMD is the only sticker easy to see from the outside of the box.

Price = quality in many decision maker's minds. Are you willing to put your job and reputation on the line for it to buy 20,000 computers? This has nothing to do with the quality of AMD's chips. It has all to do with the psychology of people. This isn't even limited to AMD. It is running a business 101.
http://www.marketingdonut.co.uk/sales/sales-techniques-and-negotiations/don-t-compete-on-price
https://creativepro.com/three-big-reasons-never-compete-price/
http://www.nextmarketing.com.au/201...-3-reasons-you-should-never-compete-on-price/
http://consultantjournal.com/blog/why-you-never-want-to-compete-on-price

I think with companies like Baidu and Microsoft being initial customers for Epyc it will help AMD in that regard as they have significant resources. The Baidu CEO was confident enough to talk about the product on behalf of AMD for example.

We all know Intel will still outsell AMD,but its about AMD carving out its own little pocket in the higher margin markets - Intel by its very size has much larger overheads so needs to support a much larger sales base,so even if AMD makes a relatively smallish dent in Intel sales,it would be enough to improve their finances quite a bit.

 

[stockolicious]

It is interesting to see AMD and Intel swap positions here. I remember way back AMD bragging about Phenom being "native quad core" vs. the Kentsfield being a couple of dies under one heat spreader.

It'd be pretty cool to see this technology find its way into the GPU space.

I dont think about this from a technology perspective as much as i do about the cost aspect. AMD is making CPU's more of a commodity (part of a product not the product) that is only because they have now become more competitive and people will realize a CPU is a part in a computer or a server.
Intel runs commercials with their branding all the time like they are an end product but in fact they are a part that makes products - this only happened because they did a great job at getting to a monopoly position. Intel is involved in trying to grow their company in many different avenues but their monopoly in CPU's and DataCenters is over. AMD and NVDA have seen to that.
one more point - as Intel goes further into self driving cars and the like id rather not have a voice in my car come out and say this is all brought to you from "skylake inside". In the end CPU's are a commodity - not a brand - will be an interesting next 18 months. id say buy some AMD stock.

 

[VirtualLarry]

On the low end, low price is killing AMD. For decades, AMD had the cheap chips and OEMs put them into cheap computers with cheap motherboards and cheap components. Guess who gets blamed if something is slow, bad, or malfunctions? AMD is the only sticker easy to see from the outside of the box.

Too true. I've seen plenty of failing AMD-based OEM rigs that failed pre-maturely, compared to comparable Intel rigs, precisely because the OEMs belived that AMD was the "budget" brand, and therefore they cut quality corners, to build to a price point.

 

[Soulkeeper]

What is the core config on Ryzen 1900X ?
Is it two dies with a ccx disabled on each, or is it a single die. If the later then how do they manage quad channel memory ?

 

[tamz_msc]

What is the core config on Ryzen 1900X ?
Is it two dies with a ccx disabled on each, or is it a single die. If the later then how do they manage quad channel memory ?

It's two dies each with the configuration of the Ryzen 5 1400 and higher clocks.

 

[coercitiv]

What is the core config on Ryzen 1900X ?

Two dies, half of L3 cache is disabled. Can tell from the AMD slide bellow: memory channels stay the same, relative cache size drops.

Spoiler: AMD Slide

 

[wildhorse2k]

1900X with halved cache, essentially being dual 4 core system with a BIOS switch for creator/gaming mode sounds like a very bad idea. Useful perhaps only if you need to use those PCIe lanes for storage or GPUs, but nothing else.

 

[UglyDuckling]

1900X with halved cache, essentially being dual 4 core system with a BIOS switch for creator/gaming mode sounds like a very bad idea. Useful perhaps only if you need to use those PCIe lanes for storage or GPUs, but nothing else.

Game mode is operational on the fly from the OS.

It sounds bad, but why not put some stock into what AMD has achieved here?

Just consider the good reviews it has, it can't all be made up.

 

[french toast]

Game mode is operational on the fly from the OS.

It sounds bad, but why not put some stock into what AMD has achieved here?

Just consider the good reviews it has, it can't all be made up.

Its a conspiracy I tell thee, can't be that fast and efficient with all that glue.

 

[Guru]

You clearly don't understand the size of Intel, how much money they actually clear on CPUs, and just how small AMD is in comparison. You're simply wrong and claiming Intel is "dead" was laughable and still is.

So they are so big that they can sell 18 cores CPU's at a constant loss for a whole generation?

Their CPU's were designed to be big and expensive, its a monolithic die, as opposed to amd's smaller dies "glued" together. Intel has to SPECIFICALLY develop each processor, so 18 core, 16 core, 14 core, 12 core, 10 core, etc... are all specifically developed, they can't just put two and two together.

For intel to be selling a whole generation of CPU's at a loss would be suicide. They'd rather still sell okay'ish, but at a profit, rather than sell a lot, but at a loss.

Again we are talking about 10+ core chips here, not the low cost 2 and 4 cores they've been milking for 7+ years so far at an advanced node from the rest of the competition, Intel are also stuck at 16nm and their plans for 10nm have hit a brick wall, they are probably spending billions and billions trying to reach 10nm and still haven't achieved it. According to their slides 10nm was supposed to be ready at least a year back, now under their own guidance its mid 2018 at the soonest.

This is not monopoly, this is not your childrens card games. They can't sell their processors at a loss for a constant generation just to be competitive price wise with AMD.

 

[IndyColtsFan]

So they are so big that they can sell 18 cores CPU's at a constant loss for a whole generation?

Their CPU's were designed to be big and expensive, its a monolithic die, as opposed to amd's smaller dies "glued" together. Intel has to SPECIFICALLY develop each processor, so 18 core, 16 core, 14 core, 12 core, 10 core, etc... are all specifically developed, they can't just put two and two together.

For intel to be selling a whole generation of CPU's at a loss would be suicide. They'd rather still sell okay'ish, but at a profit, rather than sell a lot, but at a loss.

Again we are talking about 10+ core chips here, not the low cost 2 and 4 cores they've been milking for 7+ years so far at an advanced node from the rest of the competition, Intel are also stuck at 16nm and their plans for 10nm have hit a brick wall, they are probably spending billions and billions trying to reach 10nm and still haven't achieved it. According to their slides 10nm was supposed to be ready at least a year back, now under their own guidance its mid 2018 at the soonest.

This is not monopoly, this is not your childrens card games. They can't sell their processors at a loss for a constant generation just to be competitive price wise with AMD.

You're right that this isn't a child's Monopoly game, and I suggest you stop thinking like you're playing it. You do know that Intel makes HUGE margins on their chips currently, right? Do you understand what margin is? Do you understand the money Intel makes and the volume they sell? From your comments, you clearly don't and anyone who seriously thinks AMD is going to "kill" Intel because of Ryzen is ignorant of history. AMD couldn't kill Intel when they had almost 50% market share, were the clear performance leader, AND were cheaper and you think they will now? HAHAHAHA! Folks, you can't make this stuff up. Here is what you're missing:

Recorded record results in Q1 2017:
https://www.intc.com/investor-relat...eport-First-Quarter-2017-Results/default.aspx

Recorded record results in Q2 2017:
https://www.intc.com/investor-relat...port-Second-Quarter-2017-Results/default.aspx

Ignoring all of that, you're missing several other glaringly obvious points:

1) You're assuming Intel won't adapt. They will. See: Conroe. And frankly, we should ALL be hoping competition remains healthy as it just helps the consumer.
2) Corporate IT will likely be reluctant to go AMD for several quarters. This is what happened with Opteron and will likely happen even more with Epyc due to AMD's performance (or more specifically, lack thereof) for the last 10+ years. My prediction is that Ryzen Pro and AMD mobile solutions won't have significant penetration in the corporate world for several quarters and probably not ever. It is a shame because Ryzen is a great product.
3) Somehow, AMD can magically survive a decade of huge losses but somehow Intel, who isn't even posting losses in this segment yet, will die because they MIGHT have to settle for 40% margin instead of 50-60% margin until their next gen is ready?
4) Ryzen and Threadripper are fine products (I currently own 2 Ryzen boxes), but Intel still has the performance crown at each core count level, and some will pay the premium for that.

And let's be clear here - what I have issue with is your "Intel is dead" statement. You're hilariously wrong suggesting that Intel is "dead." You're making the very big mistake of assuming Intel has no recourse and will just keep on the current course and not respond to threats from AMD. Regardless, if had to, yes, Intel could absorb huge losses over several years and still survive. That won't happen, at least due to AMD and it won't even come to that, because a slight reduction in price (if volumes drop drastically) would likely reverse the tide and they'd still make a ton of money, albeit at lower margins per unit.

 

[IEC]

The Skylake-X chips are made from LCC and HCC dies. We can speculate about the defect rates, yields, etc. but it's clear that Intel wouldn't be selling them if they weren't making desirable margins.

It's silly to think the 800 lb gorilla (Intel) is going to be overthrown in a year by a resurgent AMD. Make no mistake, the Zen architecture is a huge win for AMD - but in the context of a decade of loss after loss. As @IndyColtsFan pointed out, Intel can still command premium margins despite some uncharacteristic fumbles in response to Zen.

Where it will get really interesting is Zen 2 and beyond.

 

[wildhorse2k]

Intel has to SPECIFICALLY develop each processor, so 18 core, 16 core, 14 core, 12 core, 10 core, etc...

What makes you think this statement is true? Do you have a technical article that would support your claim?

 

[ericlp]

Meanwhile, it if wasn't for AMD ThreadRipper with a 16 core CPU... Do you really think Intel would be coming out with an i9 20 Core CPU? Nope.... Doubtful...

Intel can only dream about a having a 16 core CPU in the supply line. AMD is probably already working on a 32 core design. Why not? As soon as Intel makes a 20 core CPU, AMD will be releasing its next gen CPU's. Do you really think AMD is going to just sit around? While Intel plays catchup?

The thing is, how long will it take Intel to make a 20 core? Might see it next year maybe mid summer, by that time Intel will want you to pay 1800 bucks for that??????? AMD will no doubt be releasing an even bigger core count. It will be fun to watch them battle it out.

Bottom line is, Intel could have made a 16 core cpu for the masses LONG ago... It took AMD to innovate for Intel to at least dream about it. A day late and dollar short. I'll buy my CPU's from AMD, not because they are faster, but because AMD made something that I wanted to see happen YEARS ago. I hope they come out with a 32 core 64 threads CPU next year or even possibly by the end of this year. Rip it up!

that means in the future Threadripper could even scale upwards to 32-cores and 64-threads

http://www.guru3d.com/news-story/am...lly-has-32-cores-under-that-heatspreader.html

 

[Kenmitch]

What makes you think this statement is true? Do you have a technical article that would support your claim?

I'm not really sure how the inner workings of cpu production and design work. but as an uneducated consumer it looks like AMD's approach is best. Way more flexibility as far as potential skus at least....As long as the glue holds.

 

[Timmah!]

Meanwhile, it if wasn't for AMD ThreadRipper with a 16 core CPU... Do you really think Intel would be coming out with an i9 20 Core CPU? Nope.... Doubtful...

Intel can only dream about a having a 16 core CPU in the supply line. AMD is probably already working on a 32 core design. Why not? As soon as Intel makes a 20 core CPU, AMD will be releasing its next gen CPU's. Do you really think AMD is going to just sit around? While Intel plays catchup?

The thing is, how long will it take Intel to make a 20 core? Might see it next year maybe mid summer, by that time Intel will want you to pay 1800 bucks for that??????? AMD will no doubt be releasing an even bigger core count. It will be fun to watch them battle it out.

Bottom line is, Intel could have made a 16 core cpu for the masses LONG ago... It took AMD to innovate for Intel to at least dream about it. A day late and dollar short. I'll buy my CPU's from AMD, not because they are faster, but because AMD made something that I wanted to see happen YEARS ago. I hope they come out with a 32 core 64 threads CPU next year or even possibly by the end of this year. Rip it up!

that means in the future Threadripper could even scale upwards to 32-cores and 64-threads

http://www.guru3d.com/news-story/am...lly-has-32-cores-under-that-heatspreader.html

1. Intel will release their 16 core i9 on 25th September. AMD already have 32 core CPU, although they sell it as server Epyc chip, not consumer oriented Threadripper.
2. While i agree Intel would not release their 16 core chip in the HEDT market without AMD competition (which is something many people on these boards have hard time to accept for whatever reason), you can bet there wont be 32 core Threadripper CPU any time soon. Not this year, not next year, maybe not in 2020. The core count pretty much doubled this year for consumers, such thing wont happen any time soon.

 

[urvile]

Personally I like to have a fairly beefy pc but I won't pay a premium so I can get slightly higher benchmarks. Yet I have no problem spending 7.5k on a watch and I have to service that regularly at a grand a pop.

I don't see Intel's grip slipping. Their hardware is everywhere. Like I was posting earlier they sold 75000 CPUs in one sale. If this was a watch forum Intel would be rolex not sure who AMD would be.....

 

[Markfw]

Meanwhile, it if wasn't for AMD ThreadRipper with a 16 core CPU... Do you really think Intel would be coming out with an i9 20 Core CPU? Nope.... Doubtful...

Intel can only dream about a having a 16 core CPU in the supply line. AMD is probably already working on a 32 core design. Why not? As soon as Intel makes a 20 core CPU, AMD will be releasing its next gen CPU's. Do you really think AMD is going to just sit around? While Intel plays catchup?

The thing is, how long will it take Intel to make a 20 core? Might see it next year maybe mid summer, by that time Intel will want you to pay 1800 bucks for that??????? AMD will no doubt be releasing an even bigger core count. It will be fun to watch them battle it out.

Bottom line is, Intel could have made a 16 core cpu for the masses LONG ago... It took AMD to innovate for Intel to at least dream about it. A day late and dollar short. I'll buy my CPU's from AMD, not because they are faster, but because AMD made something that I wanted to see happen YEARS ago. I hope they come out with a 32 core 64 threads CPU next year or even possibly by the end of this year. Rip it up!

that means in the future Threadripper could even scale upwards to 32-cores and 64-threads

http://www.guru3d.com/news-story/am...lly-has-32-cores-under-that-heatspreader.html

A 32 core threadripper exists today, its called EPYC. Same size chip and socket, but 32 cores, ECC not only enabled, required (I think) and setup for a 1P or 2P config.

 

[IEC]

Meanwhile, it if wasn't for AMD ThreadRipper with a 16 core CPU... Do you really think Intel would be coming out with an i9 20 Core CPU? Nope.... Doubtful...

Intel can only dream about a having a 16 core CPU in the supply line. AMD is probably already working on a 32 core design. Why not? As soon as Intel makes a 20 core CPU, AMD will be releasing its next gen CPU's. Do you really think AMD is going to just sit around? While Intel plays catchup?

The thing is, how long will it take Intel to make a 20 core? Might see it next year maybe mid summer, by that time Intel will want you to pay 1800 bucks for that??????? AMD will no doubt be releasing an even bigger core count. It will be fun to watch them battle it out.

Bottom line is, Intel could have made a 16 core cpu for the masses LONG ago... It took AMD to innovate for Intel to at least dream about it. A day late and dollar short. I'll buy my CPU's from AMD, not because they are faster, but because AMD made something that I wanted to see happen YEARS ago. I hope they come out with a 32 core 64 threads CPU next year or even possibly by the end of this year. Rip it up!

that means in the future Threadripper could even scale upwards to 32-cores and 64-threads

http://www.guru3d.com/news-story/am...lly-has-32-cores-under-that-heatspreader.html

For clarity:
1) Intel already has a 28c/56t CPU. While it costs an arm and a leg because it is a monolithic die, there's no technical reason why Intel can't release higher core counts for HEDT with a X299v2 in the future (margins aside). It's pretty clear that's what they're doing in X299 by making HCC dies available to HEDT for the first time.
Source: https://ark.intel.com/products/family/595/Intel-Xeon-Processors
2) Intel already has their version of "glue" in the works. It's simply the most economical way to scale to more cores without breaking the bank. Expect more cores in the core wars from everyone, Soon.™
3) AMD has confirmed that Threadripper is 2 dies + 2 silicon spacers. Even if it was 4 dies, X399 is quad channel, and you get 2 channels of memory from each die... there's nothing stopping AMD from releasing a Zen 2 with more cores, but for now Threadripper tops out at 16c/32t. You want more cores? Get Epyc 1P or 2P and knock yourself out with 32 or 64 cores.

 

[french toast]

What's stopping AMD from launching 32 core threadripper with quad channel only on x399?
That's a serious question not a Rhetorical point.

 

[jpiniero]

What's stopping AMD from launching 32 core threadripper with quad channel only on x399?
That's a serious question not a Rhetorical point.

I suppose they could have made it a possibility to allow x399 to work single channel per die but there is no evidence this is actually the case.

 

[krumme]

What's stopping AMD from launching 32 core threadripper with quad channel only on x399?
That's a serious question not a Rhetorical point.

What prevents them from selling a 360w tdp with 32gb 3666 x4 ram eqipped tr? (Even in a epyc board but same name)
I think they underestimate the market and the brand value of such a product. They think more logical. Not what products also is. Feelings.
But lets see. The tr launch was far different from prior launches so it seems to me they are changing.

 

[The Stilt]

Ignore.

 

[mattiasnyc]

What prevents them from selling a 360w tdp with 32gb 3666 x4 ram eqipped tr? (Even in a epyc board but same name)
I think they underestimate the market and the brand value of such a product. They think more logical. Not what products also is. Feelings.
But lets see. The tr launch was far different from prior launches so it seems to me they are changing.

Well, could it be that they're actually maximizing profits by segmenting the market, and in addition to that are managing yields?

In other words, by doing what they're doing they get as much money out each "percentile" of the yielded dies;

- Absolute high end most efficient reliable dies go to the server market, which is willing to pay for lower speeds at higher efficiency and more support...

- Next level being TR where users can get by with less support, less high cost, but lower power efficiency and higher clock speeds.

- And then below that is Ryzen 3, 5, 7. Still good dies but not as great. Sell to the masses for a lot less.

Isn't it possible that the above is the most lucrative and that whatever amount of dies they currently produce wouldn't really "stretch" into more products right now? Because it seems to me that offering a high power high frequency high core TR might infringe upon the EPYC platform, or vice versa if they turned up the speed of EPYC chips.....

Or am I thinking about it backward?

 

[Topweasel]

They can't turn up the speed on the EPYC chips. Not really. The R7 is already an extremely efficient die at 1800x considering the core count. The R1700 takes it to another level. But power usage as you see on TR is pretty linear. The 1950x is roughly 2x 1800. Stands to reason that the same would apply to an EPYC. 4x ~90w usage would be 360w. 4x 65w would be 260w.

As we know with the 1700 it generally doesn't clock as high as a 1800x. So chances are they aren't really competing in dies. A TR die is about clocking as high as possible within a power envelope. An EPYC die is about hitting a pretty low clockspeed with as low a power usage as possible. A 7601 for example uses the same 180w. That means for it's clockspeed it needs to limit power usage to 45w a die vs. 90w on TR. They are basically on the competing ends of the spectrum. EPYC is also not going to see a jump in power allowance since cooling is more restricted in the server space than desktop and even then AMD would need to open it up to an insane 360w for the die choices to compete.