The AMD Threadripper 2 Teaser: Pre-Orders Start Today, Up to 32 Cores
by Ian Cutress on August 6, 2018 9:00 AM EST_678x452.jpg)
The biggest news to come out of Computex, AMD’s second generation of its Ryzen Threadripper platform, is almost here. Today’s announcement is all amount images, speeds and feeds, specifications, and an ‘unboxing’ announcement, leading to pre-orders a week before retail. As much as it pains me that there is an unboxing embargo and pre-orders before we even know how the new chips will perform, here we are. Today we get to go through the on-box specifications, discuss the design, and show what AMD included in our press kit.
2990WX Pre-Orders Open Today, Retail August 13th
The formal launch of the new parts is one week from today, and in the biggest markets the first processor of the launch should be available for pre-order. AMD is set to launch four versions of its second generation Ryzen Threadripper over the course of Q3, starting with its flagship 32-core.
| AMD SKUs | |||||||
| Cores/ Threads |
Base/ Turbo |
L3 | DRAM 1DPC |
PCIe | TDP | SRP | |
| TR 2990WX | 32/64 | 3.0/4.2 | 64 MB | 4x2933 | 60 | 250 W | $1799 |
| TR 2970WX | 24/48 | 3.0/4.2 | 64 MB | 4x2933 | 60 | 250 W | $1299 |
| TR 2950X | 16/32 | 3.5/4.4 | 32 MB | 4x2933 | 60 | 180 W | $899 |
| TR 2920X | 12/24 | 3.5/4.3 | 32 MB | 4x2933 | 60 | 180 W | $649 |
| Ryzen 7 2700X | 8/16 | 3.7/4.3 | 16 MB | 2x2933 | 16 | 105 W | $329 |
Sitting at the top of the stack is the 32-core 2990WX, which will retail at $1800, competing directly against Intel’s existing 18-core part, the Core i9-7980XE. AMD is using "WX" in the name here to offer some branding consistency with their high-end workstation graphics models, the Radeon Pro WX series.
Beating at its heart are four 12nm "Zen+" Zeppelin dies, each with eight cores, connected by AMD’s Infinity Fabric, and using simultaneous multi-threading for a total of 64 threads. As with previous generations, there are 60 PCIe 3.0 lanes for add-in cards and storage, and another four lanes for the chipset. There is a slight change in the memory speed, with the new parts supporting DDR4-2933.
Joining the WX line is the Ryzen Threadripper 2970WX, set at $1299 MSRP. This 24-core part uses three cores per CCX (so six cores per die, four dies per package), and also has simultaneous multi-threading for a total of 48 threads. Speeds are equal to the TR 2990WX, with a 3.0 GHz base clock and a 4.2 GHz single core turbo. Per-core turbo speeds will be decided by Precision Boost 2 and XFR2, as explained below. The 2970WX is set for an October launch.
Both of the WX models are hitting the new TDP of 250W, although the way AMD measures TDP is heavily dependent on the cooler used. In this case, their new high-end ‘Wraith Ripper’ cooler is the benchmark, which is developed by Cooler Master. This new cooler is very large, but offers full heatspreader coverage – which as we've discovered is critical for Threadripper cooling – and provides good clearance for memory. It will be sold separately in the channel, at around $100 MSRP.
Also on the table are the 2950X ($899) and the 2920X ($649), which use the existing X branding. This is because these processors are direct replacements for the 1950X and 1920X, using only two active dies in each chip, but this time around are using Zen+, the second generation Ryzen features such as faster caches and better frequency characteristics. The 16-core TR 2950X will have a base frequency of 3.5 GHz, a turbo of 4.4 GHz, and is due to launch on August 31st. The 12-core TR 2920X by comparison also has a 3.5 GHz base frequency, but a slightly lower turbo at 4.3 GHz, and is due out in October. Both of these parts are set at 180W TDP, like the first generation chips.
| AMD Ryzen Product Stacks & Launch Prices | |||
| Ryzen 1000 (2017) | Ryzen 2000 (2018) | ||
| - | TR 2990WX (32C) | $1799 | |
| - | TR 2970WX (24C) | $1299 | |
| TR 1950X (16C) | $999 | TR 2950X (16C) | $899 |
| TR 1920X (12C) | $749 | TR 2920X (12C) | $649 |
| TR 1900X (8C) | $599 | - | |
| Ryzen 7 1800X (8C) | $499 | Ryzen 7 2700X (8C) | $329 |
| Ryzen 7 1700X (8C) | $399 | ||
At this point AMD has not stated if it will expand the family even further, so we don't know if the 1900X will stay at the bottom of the stack, be replaced with a 2000 series model, or if it dies off completely.
At this point, AMD’s main competition is still with Intel’s Skylake-X parts. The 32-core 2990 WX will align in price with the 18-core Core i9-7980XE.
| Threadripper 2 vs Skylake-X The Battle (Sorted by Price) |
||||||||
| Cores/ Threads |
Base/ Turbo |
L3 | DRAM 1DPC |
PCIe | TDP | SRP | ||
| Intel | i9-7980XE | 18/36 | 2.6/4.4 | 24.75 | 4x2666 | 44 | 140W | $1999 |
| AMD | TR 2990WX | 32/64 | 3.0/4.2 | 64 MB | 4x2933 | 60 | 250W | $1799 |
| Intel | i9-7960X | 16/32 | 2.8/4.4 | 22.00 | 4x2666 | 44 | 140W | $1699 |
| Intel | i9-7940X | 14/28 | 3.1/4.4 | 19.25 | 4x2666 | 44 | 140W | $1399 |
| AMD | TR 2970WX | 24/48 | 3.0/4.2 | 64 MB | 4x2933 | 60 | 250W | $1299 |
| Intel | i9-7920X | 12/24 | 2.9/4.4 | 16.50 | 4x2666 | 44 | 140W | $1199 |
| Intel | i9-7900X | 10/20 | 3.3/4.3 | 13.75 | 4x2666 | 44 | 140W | $980 |
| AMD | TR 2950X | 16/32 | 3.5/4.4 | 32 MB | 4x2933 | 60 | 180W | $899 |
| AMD | TR 2920X | 12/24 | 3.5/4.3 | 32 MB | 4x2933 | 60 | 180W | $649 |
| Intel | i7-7820X | 8/16 | 3.6/4.3 | 11 MB | 4x2666 | 28 | 140W | $593 |
| AMD | TR 1900X | 8/16 | 3.8/4.0 | 16 MB | 4x2666 | 60 | 180W | $549 |
| AMD | R7 2700X | 8/16 | 3.7/4.3 | 16 MB | 2x2933 | 16 | 105W | $329 |
Across the metrics, AMD has more cores, is competitive on frequency, has more memory channels, more PCIe lanes, and supports higher memory clockspeeds. Intel has lower power, and above 16 cores a slight turbo clockspeed advantage. Meanwhile we already know from Ryzen 2000-series testing that cache speeds are a back-and-forth competition, and Intel has a slight IPC advantage. Game on.
Facebook
Twitter
RSS
101 Comments
View All Comments
SetiroN - Monday, August 6, 2018 - link
The memory configuration is going to be a huge bottleneck.Just try you try to use a 32 core Epyc with only 4 channels populated: performance it's hindered so badly you end up making very little use of the additional cores unless you're not accessing memory at all.
This all feels like an afterthought.
artk2219 - Monday, August 6, 2018 - link
So you're telling me AMD is shaving off features from their more expensive server parts so that theres some market differentiation? For shame! Seriously though, it is annoying that TR4 and SP3 are "2 different sockets", would have been nice to be able to use Epyc's in TR4.drajitshnew - Monday, August 6, 2018 - link
My "guess" is that while TR4 ( SP3R2) and SP3 are both 4094 pins, in TR4 the pins leading to the 2nd 2 processors are just that-- pins. They are just for physical support & are not electrically connected to anything. Hence, to maintain backwards comptibility AMD disabled the memory & PCIE of the second pair of dieseastcoast_pete - Monday, August 6, 2018 - link
While I also believe that there is no such thing as too much computing power, the 32 (and 24?) core TRs are the CPU equivalents of a 1,000 HP engine in a car: great for bragging rights, but only useful in very specific situations, and otherwise not faster than mere 8 core chips. In this case, the applications where 32 cores can make a difference are those that are not that dependent on memory speed/access. I would love to see some benchmarks for compiling and complex CAD situations.Overall, the question is/remains how well AMD executed on this second round of "NUMA on a chip".
Lastly, about EPYC vs. TR: AMD learned from the master (Intel). It's not about not letting people run server chips on desktop boards, it's about blocking people from doing the opposite: using much less expensive desktop CPUs in server boards and for server applications. That is also why desktop CPUs and chipsets basically never support ECC RAM, which is a requirement for many servers. TR is almost "EPYC", but just not quite, so you still have to buy EPYC and pay epic prizes for your servers. But than, Intel does the same, and gouges us even worse.
mapesdhs - Monday, August 6, 2018 - link
Not sure how these are about blocking people from doing the opposite, since they do support ECC, so surely one could use these CPUs just as they are with a good quality consumer mbd and they'd do just fine for a wide range of server tasks, using ECC memory if desired. If companies cared about cost that much then this is an option. Most though won't do that. There's a belief that companies will cram a consumer chip onto a pro board if they can, but really that's very rare as most bulk buyers of workstations and servers get them from OEMs, very few build their own.Nobody's gouging anyone btw, it's still a free market choice whether to buy Intel or not.
smilingcrow - Monday, August 6, 2018 - link
In theory TR boards can support ECC but I've heard reports that validation of ECC RAM is not exactly a priority and with all the work Ryzen boards required regarding RAM that's not a surprise.So anybody here built a TR ECC system and how did you get on? 1st hand reports are always better.
Oxford Guy - Tuesday, August 7, 2018 - link
ECC RAM is sold at slower speeds than typical enthusiast RAM. I fail to see why validation would be necessary. The fastest ECC RAM I know of is only 2666. If there is anything faster it should still fit within the TR2 spec.imaheadcase - Monday, August 6, 2018 - link
So why did the CPU race slow to a crawl now for years? Have we actually reached a "safe" limit for CPUs until some new tech can make it faster? I know the need isn't as great as it used to be, but remember the days that CPU speed leaped so much each generation..like 500mhz jumps each new CPU it seemed. Now we are seeing boosts..which is basically like saying "We can go this high, but its just a limit because we not sure of ourselves".DigitalFreak - Monday, August 6, 2018 - link
Two reasons come to mind - technology and competition. It's becoming increasingly difficult to go to smaller process nodes (see Intel 10nm) which are necessary to make faster chips. As to competition, Intel hasn't had any until AMD's Zen architecture. They're not going to put a lot of money into R&D if they don't have to. Unfortunately for them, AMD caught them with their pants down, and their 10nm process has had nothing but problems.DigitalFreak - Monday, August 6, 2018 - link
*which are necessary to make faster chipsFaster chips without crazy heat output and power requirements, or huge die sizes.