A German retailer jumped the gun and posted a price for the mysterious new 2990X processor whose existence AMD has yet to confirm or deny. At around $1750 USD, converting from the posted €1,509 it would be significantly more expensive than the current 1950X but lower than the $2000 price tag attached to Intel's Core i9-7980XE. The price will likely actually be lower in North America as prices depend on a variety of geographically dependent charges, though it will still be a fair chunk of change. The 32 core chip is likely clocked at 3.4GHz base with a boost of 4GHz and will surpass Intel's unreleased 28 core chip in a variety of tasks and leave the i9-7980XE in the dirt with applications which prefer multiple threads, not to mention PCIe lanes.
The Inquirer links to the original leak here.
"Videocardz spotted the since-removed listing at Cyperport, which listed the 32-core CPU with a €1,509 (around £1,300) price-tag, making it roughly €500 more expensive than AMD's 16-core Threadripper 1950X at launch."
Here is some more Tech News from around the web:
- Apple and Samsung's patent feud is finally over after seven long years @ The Inquirer
- Gigabyte graphics card shipments to fall 20% in 2Q18 @ DigiTimes
- Not OK Google: Massive outage turns smart home kit utterly dumb @ The Register
- Scientists Develop Thermal Camouflage That Can Dupe Infrared Cameras @ Slashdot
- Ticketmaster was warned about data breach two months ago, Monzo claims @ The Inquirer
- Why, hello Rubrik's Trello: Data protection biz leaves productivity tool open to world+dog @ The Register
- Every Android Device Launched Since 2012 Impacted By RAMpage Vulnerability @ Slashdot
- Google kills AdWords! @ The Register
- Zyxel Multy X AC3000 WiFi Mesh System @ Benchmark Reviews
“Videocardz spotted the
“Videocardz spotted the since-removed listing at Cyperport, which listed the 32-core CPU with a €1,509 (around £1,300) price-tag, making it roughly €500 more expensive than AMD’s 16-core Threadripper 1950X at launch.”
Let’s assumed the leak is true… the 32 core is $500 more than the 16 cores. What’s the problem, here… again??? Are we expecting they priced the same as the TR 1950X??? Seriously… I meant seriously????
AMD are not allowed to make money..AT ALL!
*Sigh*
With four active dies in the
With four active dies in the 2990X, one would reasonably hope for it to cost four times what the top end single die SKU did (320 USD -> 1280 USD). This would be in line with how the 1950X with two active dies debuted at exactly two times what the top end single die SKU from its generation did (500 USD -> 1000 USD).
The possible 1750 USD price point of the 2990X is well in excess of that. People might justifiably be miffed about it.
Perhaps the more relevant criticism here would be how the pricing of the first generation Threadripper processors hasn’t kept proper scale as the prices for its single die brethren have fallen.
Then please, explained to me
Then please, explained to me how Intel pricing according to your logic.
Yeahhh… Like I said… AMD might as well file for bankruptcy
I never had any expectation
I never had any expectation that there would be a linear scale with die count. That doesn’t happen with much of anything else. Higher core count processors tend to be more like exponentially more expensive. It makes sense for that to be the case. Higher end parts always come at a big price premium and high core count single die are much more expensive to make due to lower yield. There is also binning and validation, which can be quite expensive. These have to be validated differently since single die Ryzen parts don’t need to have any of the inter-package links functional. There are extra cost associated with placing the 4 die into one package in addition to the high-end price premium. It is the same die all the way from a Ryzen 3 1200 at $109 (1 die) up to a 32 core Epyc 7601 (4 die) at $4610 (current Newegg price). There is a big difference in binning, validation, and mark-up obviously. The bonding to the package processing step is also subject to yields.
This isn’t bad news at all since it is a great price, if true.
I don’t actually think that’s
I don’t actually think that’s reasonable at all. It ignores additional costs of binning and packaging – that’s not a small socket, for instance!
I’m seeing a lot of bad
I’m seeing a lot of bad counter-arguments regarding extra cost for packaging, scaling, and for some reason whatever Intel is doing.
Regarding what Intel is doing: It doesn’t matter.
Regarding scaling and packaging costs: Your argument doesn’t really pass the “it already happened” test where AMD priced according to how I proposed with last year’s 1950X. They could do it again now. It would stand to reason that it is cheaper to make a single TR4 package than it is to make four AM4 packages anyway.
Regarding HEDT premium price point: Sure, that’s reasonable. AMD already showed last year that they don’t need it though. That was the point.
And the bad news are?
And the bad news are?
VAT in EU (19% in Germany).
VAT in EU (19% in Germany).
It has AMD written on it and
It has AMD written on it and not intel, so it will never be a good CPU for PCPER.
Cause if you game in conditions you never game in, it might be slower in a way undetectable for the naked eye.
That it will be over $1000?
That it will be over $1000? Kind of thought that was pretty obvious.
The price says VAT included
The price says VAT included in cost. VAT is 19% in Germany. So no VAT included is about €1222.30. Seems to convert to about $1410.00 US.
The high end always has
The high end always has non-linear price scaling. This actually seems cheap to me considering the price of intel parts. What is the best Xeon you could get for that price? It has been said that the 28 core Intel Xeon was around $10000. The intel 18 core HEDT part is $1880 at Newegg right now, so $1500 to $1750 or so seems like a great price for 32 cores. That is almost 2x the cores for a cheaper price. It also seems like something Intel cannot compete with unless they want to lose a lot of money. The cost to make a single die 28 core part is much higher than 4 small cores; they estimate 70% more in the slides below. They would also lose money in lost sales of Xeon parts if they sold a HEDT part at a much lower price.
https://www.slideshare.net/mobile/AMD/isscc-2018-zeppelin-an-soc-for-multichip-architectures
And the First generation TR
And the First generation TR SKUs are being priced to move to make room for TR2. Intel’s high margin mark-ups have had to go lower and will have to go even lower when faced with TR2’s value as AMD continues to offer the better price/core metrics.
Zen+/TR2 and later with Zen2 and TR3 will make even more of an impact with possibily any Zen2 derived parts for the Ryzen 3rd generation SKUs that may just have some more than 8 core AM4 options in 2019, depending on what AMD does with its Epyc/Rome and 12 or 16 cores to a DIE rumored offerings. Zen2 and Epyc/Rome starting at 7nm with plenty of extra features for Zen2/TR3 and Ryzen 3rd generation in 2019.
AMD sure is delivering on its yearly improvements and Zen+/TR2 getting up to 32 cores at a relatively affordable cost/core compared to Intel’s costly cores and new Intel securiy issues on top of the meltdown issues that cost performance to mitigate against.
So Let’s subtract the VAT and convert those figures to US dollar amounts and figure on the Cost/Core that’s a much better value metric on TR, TR2 or Ryzen/Ryzen 2 and AMD’s integrated Vega graphics is sure a lot better than what Intel currently offers on its Intel/In-House Integrated Graphics based offerings.
Comparing it to the 18 core,
Comparing it to the 18 core, $2000 i9-7980XE is ridiculous. I still can’t see the bad news, even if it was $2000 it would be good news. Yes, free would be amazing…
I am starting to feel sorry
I am starting to feel sorry for Intel, old softie that I am.
AMD can wait and see what
AMD can wait and see what intel drop their 18 core to, and plonk a 24 core Epyc right next to it.
Pro Intel bias is strong with
Pro Intel bias is strong with this one. Literally only good news (what an amazing value!) yet they try to spin it as bad news.
So we’re onboard with $1000+
So we're onboard with $1000+ CPUs now, and preferring to spend less is bias? Nifty, must have missed that memo.
You would have a point if you
You would have a point if you couldn’t buy CPU’s for under $1k…except you can, and there have been many examples of >$1k processors over the past decade. If the existence of items outside your price range offends you, then I imagine your life must be very difficult.
ROFL! Show me cheaper 32c/64t
ROFL! Show me cheaper 32c/64t x86 cpu. Intel doesn’t even have one in the same segment! If you are bitching about $1500 32c64t AMD, how about 18c/36t Intel cpu that is actually more expensive ($1880 on newegg right now) in the same segment and way less cores? Why not focusing on this instead? If you prefer to spend less there are different market segments…
I don’t seem to be the one
I don't seem to be the one bitching or focusing on a single word in the title of a post about a rumour … but please don't let that stop you posting more comments.
Who knows whose nda they’ve
Who knows whose nda they’ve sign that could potentially prevent future review samples and technical knowledge should they word their article unfavourably towards certain parties.
NDAs hell, it’s those free
NDAs hell, it’s those free samples and ad revenues arrangements that speaks loudest and that does not require any sorts of NDAs, longer term or not, to put the kibosh on the whole Truth getting out there.
Really the review sites that purchase their own samples outside of any NDAs are the only ones that can be really fully trusted. It’s those page hits and free samples with NDAs attatched that make things suspect anyways and final judgements really need to be reserved for a time when the product has been relased to the general market for a few months.
It’s better to crowd fund as many reviewers directly to purchase their own samples(Outside Of NDAs) with the folks providing the funding asking the reviewers to do monthly/quarterly review updates of the samples to track any improvements/regressions from firmware, drivers, and game engine tweaks targeting the specific hardware and latest graphics APIs.
Most of the websites out there are not expending the efforts to regularly rebenchmark over a longet term to track a GPU card’s improvement/regressions as time passes.
There appears to be a lack of Website interest in the latest Graphics APIs’ ability to manage multi-GPUs on their own via the New Graphics APIs and Not CF/SLI. And that appears to be because the largest of the GPU makers is not pushing that multi-GPU path while the other maker appears to be the one that spearheaded the New Graphics API improvements that alow for Mulit-GPU Adaptor outside of the Old CF/SLI Driver methods that were not very good at that Multi-GPU task.
I just bought 1950x for
I just bought 1950x for little under 600€ (incl. taxes). Hopefully 2990x drops price in similar manner :>