5 ms·

> T1 is dead slow on floating point, the situation is so horrible to the extent that it is just a fraud, period. I mean why would anyone with a reasonable mind

by jfim 3y ago

> T1 is dead slow on floating point, the situation is so horrible to the extent that it is just a fraud, period. I mean why would anyone with a reasonable mind expect their fancy 5 figures purchase to be so slow on floating point? It was designed for the heavily parallel web workloads of the 2000s, which don't use floating point, hence the multiple threads per FPU.

tw1984 3y ago
> It was designed for the heavily parallel web workloads of the 2000s No. According to its official datasheet, "this processor targets commercial applications such as application servers and database servers". I thought floating point is an essential part of databases. That being said, given it is a fraud from the beginning, I can accept the fact that the official T1 datasheet is intentionally misleading its potential buyers. > which don't use floating point, hence the multiple threads per FPU. are you implying that it was mistake to implement more FPUs in their subsequent T2? there are certain shortcuts you don't make in engineering. when building an generic processor aiming to take on the application & database servers, you don't stupidly have a single FPU for the entire processor when letting your big mouth PR department talking about how great is the throughput.
- jfim 3y ago
  > I thought floating point is an essential part of databases. Have you ever implemented a database? Most operations (filtering, joining, indexing) are purely integer workloads with large working sets that are likely to stall the processor while waiting for data. That's exactly the scenario that was targeted with that processor design, many threads per core such that the cores are always busy even if some of the threads are stalled on memory accesses. It wasn't meant for single threaded workloads or HPC workloads, and afaik wasn't marketed for those segments.
  - tw1984 3y ago
    > Have you ever implemented a database? Yes, multiple times, in both American companies and in Chinese startups. > Most operations (filtering, joining, indexing) are purely integer workloads with large working sets that are likely to stall the processor while waiting for data. That is not a justification for having only one FPU. Let's use logic here, it is not that hard. > That's exactly the scenario that was targeted with that processor design, many threads per core such that the cores are always busy even if some of the threads are stalled on memory accesses. Again, that doesn't justify the stupidity of having just one FPU. > It wasn't meant for single threaded workloads or HPC workloads, and afaik wasn't marketed for those segments. I am now totally confused. You are defending the stupid T1 design which only has 1 FPU for its entire processor capable of running 32 threads in parallel, yet your argument is that such design is meant for multi-threaded workloads? I thought you need multiple FPUs to truly parallelize those multi-threaded workloads.
    - aleph_minus_one 3y ago
      As jfim wrote (emphasis mine): "It was designed for the heavily parallel web workloads of the 2000s". At this time, this mostly meant integer workload. For todays's workloads even in this particular domain, having only one FPU for 32 threads is likely a bad idea, but at that time, this was a perfectly acceptable compromise for this kind of application. These were simply different times.
- deaddodo 3y ago
  > I thought floating point is an essential part of databases. It's not. It's not even particularly necessary for 95% of DB Ops. I have no idea where you even gathered that thought. This drastically misinformed aspect just makes the rest of your argument seem equally misinformed.
  - tw1984 3y ago
    > It's not even particularly necessary for 95% of DB Ops. Let's say I have databases full of some sensor data that need to be processed by some fancy SQL scripts to crunch those numbers. Tell me how I am suppose to get that done without using floating point. With only one FPU in mind, tell me how to parallelize such processing and full utilize all those cores.
    - MichaelZuo 3y ago
      Picking an example from the remaining 5% does not contradict the parent's point.
      - tw1984 3y ago
        The point here is that it is pure stupidity on display to cut the number of FPUs to just one, the short lived history of such design is a proof for that.
- _a_a_a_ 3y ago
  > I thought floating point is an essential part of databases. Good heavens, no. Why on earth would you even think that, have you any experience with databases?
- speed_spread 3y ago
  > thought floating point is an essential part of databases To add to the other replies to this, the number of areas where CPU floating point performance matters is shrinking. GPUs are where it's at nowadays. Also, T1 was never going to be a gaming machine or a supercomputer node. Whoever thought otherwise didn't read the actual specs, which were very clear back then.

by jfim 3y ago

tw1984 3y ago

> It was designed for the heavily parallel web workloads of the 2000s No. According to its official datasheet, "this processor targets commercial applications such as application servers and database servers". I thought floating point is an essential part of databases. That being said, given it is a fraud from the beginning, I can accept the fact that the official T1 datasheet is intentionally misleading its potential buyers. > which don't use floating point, hence the multiple threads per FPU. are you implying that it was mistake to implement more FPUs in their subsequent T2? there are certain shortcuts you don't make in engineering. when building an generic processor aiming to take on the application & database servers, you don't stupidly have a single FPU for the entire processor when letting your big mouth PR department talking about how great is the throughput.

jfim 3y ago
> I thought floating point is an essential part of databases. Have you ever implemented a database? Most operations (filtering, joining, indexing) are purely integer workloads with large working sets that are likely to stall the processor while waiting for data. That's exactly the scenario that was targeted with that processor design, many threads per core such that the cores are always busy even if some of the threads are stalled on memory accesses. It wasn't meant for single threaded workloads or HPC workloads, and afaik wasn't marketed for those segments.
- tw1984 3y ago
  > Have you ever implemented a database? Yes, multiple times, in both American companies and in Chinese startups. > Most operations (filtering, joining, indexing) are purely integer workloads with large working sets that are likely to stall the processor while waiting for data. That is not a justification for having only one FPU. Let's use logic here, it is not that hard. > That's exactly the scenario that was targeted with that processor design, many threads per core such that the cores are always busy even if some of the threads are stalled on memory accesses. Again, that doesn't justify the stupidity of having just one FPU. > It wasn't meant for single threaded workloads or HPC workloads, and afaik wasn't marketed for those segments. I am now totally confused. You are defending the stupid T1 design which only has 1 FPU for its entire processor capable of running 32 threads in parallel, yet your argument is that such design is meant for multi-threaded workloads? I thought you need multiple FPUs to truly parallelize those multi-threaded workloads.
  - aleph_minus_one 3y ago
    As jfim wrote (emphasis mine): "It was designed for the heavily parallel web workloads of the 2000s". At this time, this mostly meant integer workload. For todays's workloads even in this particular domain, having only one FPU for 32 threads is likely a bad idea, but at that time, this was a perfectly acceptable compromise for this kind of application. These were simply different times.
deaddodo 3y ago
> I thought floating point is an essential part of databases. It's not. It's not even particularly necessary for 95% of DB Ops. I have no idea where you even gathered that thought. This drastically misinformed aspect just makes the rest of your argument seem equally misinformed.
- tw1984 3y ago
  > It's not even particularly necessary for 95% of DB Ops. Let's say I have databases full of some sensor data that need to be processed by some fancy SQL scripts to crunch those numbers. Tell me how I am suppose to get that done without using floating point. With only one FPU in mind, tell me how to parallelize such processing and full utilize all those cores.
  - MichaelZuo 3y ago
    Picking an example from the remaining 5% does not contradict the parent's point.
    - tw1984 3y ago
      The point here is that it is pure stupidity on display to cut the number of FPUs to just one, the short lived history of such design is a proof for that.
_a_a_a_ 3y ago
> I thought floating point is an essential part of databases. Good heavens, no. Why on earth would you even think that, have you any experience with databases?
speed_spread 3y ago
> thought floating point is an essential part of databases To add to the other replies to this, the number of areas where CPU floating point performance matters is shrinking. GPUs are where it's at nowadays. Also, T1 was never going to be a gaming machine or a supercomputer node. Whoever thought otherwise didn't read the actual specs, which were very clear back then.