3 ms·

> People who bought into the AM4 ecosystem 5 years ago should be able to update to these brand new zen3 cpus. As long as people recognise thay a brand new CPU

> People who bought into the AM4 ecosystem 5 years ago should be able to update to these brand new zen3 cpus. As long as people recognise thay a brand new CPU in a 5 year old board will have some compromises... Which given the rates of #internetdrama these days, is probably not likely.

kemotep 5y ago
It reduces potential e-waste but depending on the board there certainly will be issues getting the best performance out of the newer zen 3 cpus. No pcie 4.0 support being a big one.
- belval 5y ago
  I see this line repeated a lot, but in practice is there any hard data that PCIe 3.0 is bottlenecking reasonably priced hardware? Doesn't seem like it (RTX 3090 is probably at the edge of what I'd consider "reasonably priced" anyway): https://www.reddit.com/r/hardware/comments/ikrteg/nvidia_rep_on_pcie_30_vs_40_for_the_rtx_3090_the/ https://www.reddit.com/r/hardware/comments/ikrteg/nvidia_rep...
  - bcrosby95 5y ago
    I wouldn't call a $1400 graphics card anywhere close to "reasonably priced". $400 maybe.
    - belval 5y ago
      Right, but that's my point, if PICe 3.0 is not a significant bottleneck to an RTX 3090, it can hardly be seen as one for anyone running non-enthusiast builds.
  - kmeisthax 5y ago
    Most games can run fine on Thunderbolt external GPUs, and that's a PCIe 3.0x4 link (roughly) that, on most Intel implementations, is also bottlenecked by the chipset[0]. Game assets are typically static, so available bandwidth mainly impacts how fast you can stream data into VRAM. Of course, with things like DirectStorage and consoles all embracing SSD-to-GPU streaming and hardware decompression, game developers might actually bother to actually use the extra bandwidth they've been given. However, that's still a way's off and definitely not a requirement to enjoy most games today. The reason why PCIe 4.0 actually became a thing was because of enterprise storage arrays. M.2 slots and U.2 connectors don't have enough pins for 16 lanes, and using up so many of those lanes for one device makes no sense if you need to stick 10 or 20 of them in a server. That's also a use case that doesn't really make sense on AM4, unless you have a bifurcation[1]-capable motherboard or are spending way too much money on M.2 carriers with PLX chips in them. [0] AFAIK, Intel wanted Thunderbolt direct-to-CPU but there was some weird driver/certification nonsense with Microsoft or something, and going through the chipset apparently made it easier for vendors not named Apple to support it. I don't remember the details. [1] The ability to drive multiple PCIe devices off the same slot by splitting the slot's lanes. Most M.2 carrier boards are wired up to work this way because proper PCIe hubs are absurdly expensive... because the entire market for such chips are just storage array vendors.
  - paulmd 5y ago
    Not generally, but there are a few edge-cases. AMD recently released the 6500XT GPU which has PCIe 4.0 but only x4, and when used in PCIe 3.0 motherboards that obviously becomes 3.0x4. Furthermore, AMD only put 4GB of VRAM on it, which isn't good enough for modern games (it's actually less than the previous 5500XT, where AMD made a big advertising push about "4GB isn't enough for modern games anymore") so it swaps all the time, which amplifies the PCIe bottleneck. The card loses something like 15% performance when used on a PCIe 3.0 motherboard. It's not like it won't run, but that's a significant amount of performance, that's like a half a tier of performance and far more than you see on (eg) 3090. Unfortunately we're still in something of a transition period. AMD blocked vendors from supporting PCIe 4.0 on 300/400 series boards that might have been capable of the required signal integrity (particularly on the first slot). AMD doesn't support PCIe 4.0 on their Zen3 APUs at all either - and some of the new processors are based on the APU die even though they don't have graphics, so they are limited to PCIe 3.0 as well. And obviously Skylake/Coffee Lake/Comet Lake stuff is all PCIe 3.0 based since they're ancient. So there are definitely scenarios where you might think "throw in a cheap dGPU" and are still stuck with PCIe 3.0. Anyway though, what I would caution you here is, the 3090 has lots of VRAM, so it doesn't swap. What the 6500XT shows is, low-end cards can be more susceptible to PCIe bottlenecking - they have less VRAM, so they swap more, which increases the pressure on the PCIe bus. 3090 results are not representative of a worst case scenario just because they do drawcalls really fast (high framerate), there are other parts of the pipeline where PCIe load can be generated. If you are swapping due to low VRAM, that's still PCIe load. Similarly - simply using a card placed into a PCIe 3.0x4 slot is not representative of Thunderbolt 3 results either, despite both links being 3.0x4. Thunderbolt is usually implemented as an additional standalone chip attached to the chipset - so it's not a CPU-direct link, there's multiple hops with higher latency there. There is also contention for the chipset bandwidth - the chipset has a 3.0x4 link, so the GPU alone can saturate it, but there's also NVMe traffic (particularly for pre-11th gen Intel where NVMe has to attach to the chipset) and network traffic (chipset provides the network controller), etc. It's bidirectional so you can read at 4GB/s while you write 4GB/s to the GPU, but there's also just general contention for commands/etc. So performance results on Thunderbolt will be worse than a card attached to the chipset, which will be worse than a card attached to CPU-direct PEG lanes, even if lane count is the same. (the exception to this might be Ice Lake and newer Intel laptop chips, where the Thunderbolt controller is actually part of the CPU itself, the performance impact of that should be less. However, this does not apply to desktop chips, including Alder Lake.)
  - eulers_secret 5y ago
    PCIe 3.0 is a bottleneck for modern SSDs. It's not a big deal at all, but it is a bottleneck.
  - matja 5y ago
    $100 USD PCIe 4.0 SSDs exist (e.g. 500GB SN850), which can comfortably run at close to the theoretical limit of PCIe 4.0, bottle-necked on PCIe 3.0.
    - deckard1 5y ago
      yeah but what's the use case here? I've been on NVMe since 2017. Without being told about it, I would never know it's not a regular SSD. PCIe 4.0 is one of those things that if you need it, then you already know you need it. But almost no one does.
jotm 5y ago
Only thing I can think of is RAM/PCIe speed... which isn't much of an issue. What else is there?

kemotep 5y ago

It reduces potential e-waste but depending on the board there certainly will be issues getting the best performance out of the newer zen 3 cpus. No pcie 4.0 support being a big one.

belval 5y ago
I see this line repeated a lot, but in practice is there any hard data that PCIe 3.0 is bottlenecking reasonably priced hardware? Doesn't seem like it (RTX 3090 is probably at the edge of what I'd consider "reasonably priced" anyway): https://www.reddit.com/r/hardware/comments/ikrteg/nvidia_rep_on_pcie_30_vs_40_for_the_rtx_3090_the/ https://www.reddit.com/r/hardware/comments/ikrteg/nvidia_rep...
- bcrosby95 5y ago
  I wouldn't call a $1400 graphics card anywhere close to "reasonably priced". $400 maybe.
  - belval 5y ago
    Right, but that's my point, if PICe 3.0 is not a significant bottleneck to an RTX 3090, it can hardly be seen as one for anyone running non-enthusiast builds.
- kmeisthax 5y ago
  Most games can run fine on Thunderbolt external GPUs, and that's a PCIe 3.0x4 link (roughly) that, on most Intel implementations, is also bottlenecked by the chipset[0]. Game assets are typically static, so available bandwidth mainly impacts how fast you can stream data into VRAM. Of course, with things like DirectStorage and consoles all embracing SSD-to-GPU streaming and hardware decompression, game developers might actually bother to actually use the extra bandwidth they've been given. However, that's still a way's off and definitely not a requirement to enjoy most games today. The reason why PCIe 4.0 actually became a thing was because of enterprise storage arrays. M.2 slots and U.2 connectors don't have enough pins for 16 lanes, and using up so many of those lanes for one device makes no sense if you need to stick 10 or 20 of them in a server. That's also a use case that doesn't really make sense on AM4, unless you have a bifurcation[1]-capable motherboard or are spending way too much money on M.2 carriers with PLX chips in them. [0] AFAIK, Intel wanted Thunderbolt direct-to-CPU but there was some weird driver/certification nonsense with Microsoft or something, and going through the chipset apparently made it easier for vendors not named Apple to support it. I don't remember the details. [1] The ability to drive multiple PCIe devices off the same slot by splitting the slot's lanes. Most M.2 carrier boards are wired up to work this way because proper PCIe hubs are absurdly expensive... because the entire market for such chips are just storage array vendors.
- paulmd 5y ago
  Not generally, but there are a few edge-cases. AMD recently released the 6500XT GPU which has PCIe 4.0 but only x4, and when used in PCIe 3.0 motherboards that obviously becomes 3.0x4. Furthermore, AMD only put 4GB of VRAM on it, which isn't good enough for modern games (it's actually less than the previous 5500XT, where AMD made a big advertising push about "4GB isn't enough for modern games anymore") so it swaps all the time, which amplifies the PCIe bottleneck. The card loses something like 15% performance when used on a PCIe 3.0 motherboard. It's not like it won't run, but that's a significant amount of performance, that's like a half a tier of performance and far more than you see on (eg) 3090. Unfortunately we're still in something of a transition period. AMD blocked vendors from supporting PCIe 4.0 on 300/400 series boards that might have been capable of the required signal integrity (particularly on the first slot). AMD doesn't support PCIe 4.0 on their Zen3 APUs at all either - and some of the new processors are based on the APU die even though they don't have graphics, so they are limited to PCIe 3.0 as well. And obviously Skylake/Coffee Lake/Comet Lake stuff is all PCIe 3.0 based since they're ancient. So there are definitely scenarios where you might think "throw in a cheap dGPU" and are still stuck with PCIe 3.0. Anyway though, what I would caution you here is, the 3090 has lots of VRAM, so it doesn't swap. What the 6500XT shows is, low-end cards can be more susceptible to PCIe bottlenecking - they have less VRAM, so they swap more, which increases the pressure on the PCIe bus. 3090 results are not representative of a worst case scenario just because they do drawcalls really fast (high framerate), there are other parts of the pipeline where PCIe load can be generated. If you are swapping due to low VRAM, that's still PCIe load. Similarly - simply using a card placed into a PCIe 3.0x4 slot is not representative of Thunderbolt 3 results either, despite both links being 3.0x4. Thunderbolt is usually implemented as an additional standalone chip attached to the chipset - so it's not a CPU-direct link, there's multiple hops with higher latency there. There is also contention for the chipset bandwidth - the chipset has a 3.0x4 link, so the GPU alone can saturate it, but there's also NVMe traffic (particularly for pre-11th gen Intel where NVMe has to attach to the chipset) and network traffic (chipset provides the network controller), etc. It's bidirectional so you can read at 4GB/s while you write 4GB/s to the GPU, but there's also just general contention for commands/etc. So performance results on Thunderbolt will be worse than a card attached to the chipset, which will be worse than a card attached to CPU-direct PEG lanes, even if lane count is the same. (the exception to this might be Ice Lake and newer Intel laptop chips, where the Thunderbolt controller is actually part of the CPU itself, the performance impact of that should be less. However, this does not apply to desktop chips, including Alder Lake.)
- eulers_secret 5y ago
  PCIe 3.0 is a bottleneck for modern SSDs. It's not a big deal at all, but it is a bottleneck.
- matja 5y ago
  $100 USD PCIe 4.0 SSDs exist (e.g. 500GB SN850), which can comfortably run at close to the theoretical limit of PCIe 4.0, bottle-necked on PCIe 3.0.
  - deckard1 5y ago
    yeah but what's the use case here? I've been on NVMe since 2017. Without being told about it, I would never know it's not a regular SSD. PCIe 4.0 is one of those things that if you need it, then you already know you need it. But almost no one does.

jotm 5y ago

Only thing I can think of is RAM/PCIe speed... which isn't much of an issue. What else is there?