3 ms·

Another very high performance effort is VRoom![0]. As of the latest update in the blog, they are at 10.3 DMips/MHz. 0. https://moonbaseotago.github.io/ https:

by snvzz 4y ago

Another very high performance effort is VRoom![0]. As of the latest update in the blog, they are at 10.3 DMips/MHz. 0. https://moonbaseotago.github.io/ https://moonbaseotago.github.io/

tyingq 4y ago
>they are at 10.3 DMips/MHz I understand Dhrystone isn't everything, but that's pretty impressive, and up there with high performance CPUs from the established players.
- gchadwick 4y ago
  'Isn't everything' is a huge understatement. It's a performance nice smoke test for a CPU in that if it isn't hitting the numbers you'd expect given the microarchitecture (issue width etc) you know there's a problem and due to the simplicity of the benchmark it shouldn't be hard to analyse and track down. Though there's way more a modern high performance CPU needs to do. It doesn't do anything meaningful to stress the memory system for instance. Hitting good Drystone numbers is stage 1 in a long process to building a modern high performance CPU.
  - tyingq 4y ago
    That's fair. Perhaps "is only an imperfect early indicator" would have been better than "isn't everything".
- codedokode 4y ago
  As I understand that is performance in a simulator, and it is unknown what delays are going to be between pipeline stages on real hardware.
  - rowanG077 4y ago
    Performance in a simulator is cycle accurate to the real world.
  - brucehoult 4y ago
    "delays" affect the MHz, not the performance per MHz. Tools for predicting MHz are fairly accurate. Also, VROOM works in FPGA not only in verilator.
rektide 4y ago
The practically live blogging of this VRoom! core is such a glorious thing to behold. Amazing work. And it's talked about!! We can see learn & follow that. One of the most unique singular experiences in the modern world, creating a high performance core, and we can read about it, see the process unfold, see the team iterate. Gobsmackingly amazing.
mdp2021 4y ago
Does anybody have a list of DMips/MHz per CPU results, for comparison?

by snvzz 4y ago

Another very high performance effort is VRoom![0]. As of the latest update in the blog, they are at 10.3 DMips/MHz. 0. https://moonbaseotago.github.io/ https://moonbaseotago.github.io/

tyingq 4y ago

>they are at 10.3 DMips/MHz I understand Dhrystone isn't everything, but that's pretty impressive, and up there with high performance CPUs from the established players.

gchadwick 4y ago
'Isn't everything' is a huge understatement. It's a performance nice smoke test for a CPU in that if it isn't hitting the numbers you'd expect given the microarchitecture (issue width etc) you know there's a problem and due to the simplicity of the benchmark it shouldn't be hard to analyse and track down. Though there's way more a modern high performance CPU needs to do. It doesn't do anything meaningful to stress the memory system for instance. Hitting good Drystone numbers is stage 1 in a long process to building a modern high performance CPU.
- tyingq 4y ago
  That's fair. Perhaps "is only an imperfect early indicator" would have been better than "isn't everything".
codedokode 4y ago
As I understand that is performance in a simulator, and it is unknown what delays are going to be between pipeline stages on real hardware.
- rowanG077 4y ago
  Performance in a simulator is cycle accurate to the real world.
- brucehoult 4y ago
  "delays" affect the MHz, not the performance per MHz. Tools for predicting MHz are fairly accurate. Also, VROOM works in FPGA not only in verilator.

rektide 4y ago

The practically live blogging of this VRoom! core is such a glorious thing to behold. Amazing work. And it's talked about!! We can see learn & follow that. One of the most unique singular experiences in the modern world, creating a high performance core, and we can read about it, see the process unfold, see the team iterate. Gobsmackingly amazing.

mdp2021 4y ago

Does anybody have a list of DMips/MHz per CPU results, for comparison?