5 ms·

I don't really see how the pricing is ridiculous though. They're still way cheaper than what Intel charged for the same cores a year ago, and they have more cor

I don't really see how the pricing is ridiculous though. They're still way cheaper than what Intel charged for the same cores a year ago, and they have more cores than Intel can even offer on their best workstation chips. If you compare against 28 core Xeons, the new Threadrippers are a downright bargain.

bitL 7y ago
I meant in comparison to what we were used to. Now a semi-decent TRX40 board is $700, entry-level TR3 is $1300. Top-end x399 board is $550, entry-level TR is $250. There was a huge jump in prices compared to previous generation of HEDT.
- barkingcat 7y ago
  This comment doesn't make sense. You're getting more cores than previous generation of HEDT, and the equivalent Intel processors aren't cheaper at all. This is a new market segment. If you want a fast cpu, the Ryzen 7 and 9 series are completely fine if you want that price range! The exact same price range as you're used to still exist. On the other hand, people have been used to paying exorbitantly for Xeon processors, like 2000-5000 per cpu, so this is a breath of fresh air.
  - bitL 7y ago
    You are getting the same number of (better) cores for higher price. But backwards/forward compatibility is gone, so you either pay the full price upfront, or are stuck with outdated chips forever.
- pitaj 7y ago
  Board cost is due to PCIe4 support mainly
  - bitL 7y ago
    Which is kinda unnecessary as there is no single GPU on the market capable of saturating PCIe3 and situations where one needs a sustained transfer between multiple M.2 SSDs that could saturate PCIe4 are very rare. Only 100Gbps+ LAN is probably practical for a few total pro users.
    - dragontamer 7y ago
      Actually, its pretty easy to get bandwidth-bottlenecked in GPU-compute. I know video games don't really get bandwidth bottlenecked, but all you gotta do is perform a "Scan" or "Reduce" on the GPU and bam, you're PCIe bottlenecked. (I recommend NVidia CUB or AMD ROCprim for these kinds of operations) CUB Device-reduce is extremely fast if the data is already on the GPU: https://nvlabs.github.io/cub/structcub_1_1_device_reduce.html https://nvlabs.github.io/cub/structcub_1_1_device_reduce.htm.... However, if the data is CPU / DDR4 RAM side, then the slow PCIe connection hampers you severely. I pushed 1GB of data to device-side reduce the other day (just playing with ROCprim), and it took ~100ms to hipMemcpy the 1GB of data to the GPU, but only 5ms to actually execute the reduce. That's a PCIe-bottleneck for sure. (Numbers from memory... I don't quite remember them exactly but that was roughly the magnitudes we're talking about). That was over PCIe 3.0 x16, which seems to only push 10GBps one-way in practice. (15Gbps in theory, but practice is always lower than the specs) Yeah, I know CPU / GPU have like 10us of latency, but you can easily write a "server" kind of CPU-master / GPU-slave scheduling algorithm to send these jobs down to the GPU. So you can write software to ignore the latency problem in many cases. Software can't solve the bandwidth problem however. You gotta just buy a bigger pipe.
  - paulmd 7y ago
    The bigger point is that there was no reason to kill X399 support at all. It's the same physical socket, the socket is capable of supporting much more than Threadripper did with it (Epyc uses the same socket for 8 memory channels), and the power consumption has not increased significantly compared to TR 2000 series. There was no reason to kill TR4. It could have been a "legacy" board with PCIe 3.0 support, like X470 is for the desktop socket. AMD just killed TR4 because they wanted everyone to buy new boards. The classic Intel move. (meanwhile Intel put a new generation of chips on X299, while also putting out a compatible X299X socket that increases lane count. Intel doing it right for once, AMD doing it wrong for once.)
- kllrnohj 7y ago
  The price jump was +8% for 24 cores and +11% for 32 cores. It wasn't really a "huge" jump. The reason it seems so much more expensive is because the cheaper SKUs were simply removed instead of replaced. And entry-level TR was never $250. That's the EOL "we need to dump old inventory" fire sale price.
  - bitL 7y ago
    Yeah, but everybody expected a price drop with new gen, instead we got overpriced TRX40 (compared to high-end x399) and the same number of cores for more $. I get it those cores are much more powerful, but it's still an untypical situation. 2990WX will probably stay at the same level as there is no other choice for x399 owners anyway.
    - kllrnohj 7y ago
      > but everybody expected a price drop with new gen No they weren't. I certainly wasn't. There was no reason at all to believe TR3 would be a price drop. Ryzen 3000 wasn't and neither was X570. If the mainstream platform parts didn't get a price drop why would the HEDT halo products? Particularly since new generations are almost never price drops, especially without any competition? > instead we got overpriced TRX40 (compared to high-end x399) X399 at launch ranged from $340 to $550. TRX40 at launch ranges from $450 to $700. Yes there was a bump there, but there is also overlap in pricing, too. You are getting PCI-E 4.0 instead along with a substantially higher spec'd chipset. You're also getting in general a higher class of board quality & construction. Similar to the X570 vs. X470 comparison. > but it's still an untypical situation Untypical in that they are actually a lot faster generation over generation, sure. Untypical in that they are priced similarly or slightly more? Not really. That's been status quo for the last decade or so. The company with the halo product sets the price. The company in 2nd place prices cuts in response. AMD has the halo, they were never going to price cut it.
      - bitL 7y ago
        Top-end TRX40 is around $1000 (Zenith II). That's almost double of Zenith Extreme x399; x399 had 16-phase VRMs as well in later releases; PCIe 4's usefulness is questionable (basically just for 100 Gigabit networking right now). For x399 users TRX40 is underwhelming as it just feels like "pay for the same stuff again" if you want to use new CPUs.
vbezhenar 7y ago
They increased pricing by $100/$200 for 24/32 core CPU since last year. May be it's not ridiculous, but they are not undercutting anymore.
- aoeusnth1 7y ago
  They are undercutting - on perf/$. The difference in sticker price you pointed out is only there because there is no comparable intel part (under $3000 at least)

bitL 7y ago

I meant in comparison to what we were used to. Now a semi-decent TRX40 board is $700, entry-level TR3 is $1300. Top-end x399 board is $550, entry-level TR is $250. There was a huge jump in prices compared to previous generation of HEDT.

barkingcat 7y ago
This comment doesn't make sense. You're getting more cores than previous generation of HEDT, and the equivalent Intel processors aren't cheaper at all. This is a new market segment. If you want a fast cpu, the Ryzen 7 and 9 series are completely fine if you want that price range! The exact same price range as you're used to still exist. On the other hand, people have been used to paying exorbitantly for Xeon processors, like 2000-5000 per cpu, so this is a breath of fresh air.
- bitL 7y ago
  You are getting the same number of (better) cores for higher price. But backwards/forward compatibility is gone, so you either pay the full price upfront, or are stuck with outdated chips forever.
pitaj 7y ago
Board cost is due to PCIe4 support mainly
- bitL 7y ago
  Which is kinda unnecessary as there is no single GPU on the market capable of saturating PCIe3 and situations where one needs a sustained transfer between multiple M.2 SSDs that could saturate PCIe4 are very rare. Only 100Gbps+ LAN is probably practical for a few total pro users.
  - dragontamer 7y ago
    Actually, its pretty easy to get bandwidth-bottlenecked in GPU-compute. I know video games don't really get bandwidth bottlenecked, but all you gotta do is perform a "Scan" or "Reduce" on the GPU and bam, you're PCIe bottlenecked. (I recommend NVidia CUB or AMD ROCprim for these kinds of operations) CUB Device-reduce is extremely fast if the data is already on the GPU: https://nvlabs.github.io/cub/structcub_1_1_device_reduce.html https://nvlabs.github.io/cub/structcub_1_1_device_reduce.htm.... However, if the data is CPU / DDR4 RAM side, then the slow PCIe connection hampers you severely. I pushed 1GB of data to device-side reduce the other day (just playing with ROCprim), and it took ~100ms to hipMemcpy the 1GB of data to the GPU, but only 5ms to actually execute the reduce. That's a PCIe-bottleneck for sure. (Numbers from memory... I don't quite remember them exactly but that was roughly the magnitudes we're talking about). That was over PCIe 3.0 x16, which seems to only push 10GBps one-way in practice. (15Gbps in theory, but practice is always lower than the specs) Yeah, I know CPU / GPU have like 10us of latency, but you can easily write a "server" kind of CPU-master / GPU-slave scheduling algorithm to send these jobs down to the GPU. So you can write software to ignore the latency problem in many cases. Software can't solve the bandwidth problem however. You gotta just buy a bigger pipe.
- paulmd 7y ago
  The bigger point is that there was no reason to kill X399 support at all. It's the same physical socket, the socket is capable of supporting much more than Threadripper did with it (Epyc uses the same socket for 8 memory channels), and the power consumption has not increased significantly compared to TR 2000 series. There was no reason to kill TR4. It could have been a "legacy" board with PCIe 3.0 support, like X470 is for the desktop socket. AMD just killed TR4 because they wanted everyone to buy new boards. The classic Intel move. (meanwhile Intel put a new generation of chips on X299, while also putting out a compatible X299X socket that increases lane count. Intel doing it right for once, AMD doing it wrong for once.)
kllrnohj 7y ago
The price jump was +8% for 24 cores and +11% for 32 cores. It wasn't really a "huge" jump. The reason it seems so much more expensive is because the cheaper SKUs were simply removed instead of replaced. And entry-level TR was never $250. That's the EOL "we need to dump old inventory" fire sale price.
- bitL 7y ago
  Yeah, but everybody expected a price drop with new gen, instead we got overpriced TRX40 (compared to high-end x399) and the same number of cores for more $. I get it those cores are much more powerful, but it's still an untypical situation. 2990WX will probably stay at the same level as there is no other choice for x399 owners anyway.
  - kllrnohj 7y ago
    > but everybody expected a price drop with new gen No they weren't. I certainly wasn't. There was no reason at all to believe TR3 would be a price drop. Ryzen 3000 wasn't and neither was X570. If the mainstream platform parts didn't get a price drop why would the HEDT halo products? Particularly since new generations are almost never price drops, especially without any competition? > instead we got overpriced TRX40 (compared to high-end x399) X399 at launch ranged from $340 to $550. TRX40 at launch ranges from $450 to $700. Yes there was a bump there, but there is also overlap in pricing, too. You are getting PCI-E 4.0 instead along with a substantially higher spec'd chipset. You're also getting in general a higher class of board quality & construction. Similar to the X570 vs. X470 comparison. > but it's still an untypical situation Untypical in that they are actually a lot faster generation over generation, sure. Untypical in that they are priced similarly or slightly more? Not really. That's been status quo for the last decade or so. The company with the halo product sets the price. The company in 2nd place prices cuts in response. AMD has the halo, they were never going to price cut it.
    - bitL 7y ago
      Top-end TRX40 is around $1000 (Zenith II). That's almost double of Zenith Extreme x399; x399 had 16-phase VRMs as well in later releases; PCIe 4's usefulness is questionable (basically just for 100 Gigabit networking right now). For x399 users TRX40 is underwhelming as it just feels like "pay for the same stuff again" if you want to use new CPUs.

vbezhenar 7y ago

They increased pricing by $100/$200 for 24/32 core CPU since last year. May be it's not ridiculous, but they are not undercutting anymore.

aoeusnth1 7y ago
They are undercutting - on perf/$. The difference in sticker price you pointed out is only there because there is no comparable intel part (under $3000 at least)