5 ms·

The cores/money comparison doesn't looks like it accounts for inflation.

by cl0ckt0wer 3y ago

huijzer 3y ago
Indeed. I was also wondering why people would buy newer chips if they don’t show performance increases per dollar.
- bee_rider 3y ago
  I think the main reason people get a new CPU, regardless of the performance, is because the device that it came in has gotten too old (wear-and tear: screen, batteries, etc).
  - rbanffy 3y ago
    I'm pretty sure that when you adjust for inflation, you get more performance per dollar, but I'd need to do math and I don't have the numbers from my head. Does Google Sheets provide a "inflation-adjusted dollar" function?
    - trashtester 3y ago
      Performace/$ is still going down when adjusting for inflation. Performance/W goes down even more.
    - bee_rider 3y ago
      I’m not sure really. But anyway, I’m not sure if it makes sense to expect performance/$ to always increase anyway. I mean, I know this started out by talking about multicore, but think about single threaded performance. They’ve already grabbed all the low hanging fruit, the challenge now is finding increasingly hard to hunt down tweaks… a small improvement might require massive engineering effort.
      - rbanffy 3y ago
        Intel may be able to squeeze a little more juice out of x86 by dropping backwards compatibility. This is something I suggested they could have done for Apple because Macs were never supposed to boot MS-DOS anyway. A simple and regular ISA is what makes ARM easier to implement, which also means lower power because of fewer transistors doing thankless work like decoding instructions and reordering them.
- pixl97 3y ago
  Because the vendor will no longer give the the old chips? That an lack of support around security.
- capitalsigma 3y ago
  There are overheads in server workloads that scale with the number of machines (network traffic, serializing/deserializing requests). There are also fixed costs per server that don't scale with core count, or at least scale sublinearly (storage, physical data center space, motherboard, ease of maintenance). So running 10 machines with 100 cores can be cheaper and more performant than running 1,000 machines with 1 core even if $/core is higher. And of course individual cores can be beefier: wider SIMD units, application-specific extensions like bfloat support for ML workloads, etc. Of course Moore's law is slowing down, but cores/$ is an extremely silly metric to use
fsckboy 3y ago
The cores/money comparison doesn't looks like it accounts for inflation.* more to the point, the comparison also represents the period of time in which AMD, the one-time "lesser" player nipping at monopolist Intel's heels by competing on price, became the technology leader and started commanding a premium for their products. Meanwhile, Intel has not been forced to cede its position in the market, their existing contracts and business model is based on premium products, not cost leadership. It will take awhile for this to sort out.

The cores/money comparison doesn't looks like it accounts for inflation.

huijzer 3y ago

Indeed. I was also wondering why people would buy newer chips if they don’t show performance increases per dollar.

bee_rider 3y ago
I think the main reason people get a new CPU, regardless of the performance, is because the device that it came in has gotten too old (wear-and tear: screen, batteries, etc).
- rbanffy 3y ago
  I'm pretty sure that when you adjust for inflation, you get more performance per dollar, but I'd need to do math and I don't have the numbers from my head. Does Google Sheets provide a "inflation-adjusted dollar" function?
  - trashtester 3y ago
    Performace/$ is still going down when adjusting for inflation. Performance/W goes down even more.
  - bee_rider 3y ago
    I’m not sure really. But anyway, I’m not sure if it makes sense to expect performance/$ to always increase anyway. I mean, I know this started out by talking about multicore, but think about single threaded performance. They’ve already grabbed all the low hanging fruit, the challenge now is finding increasingly hard to hunt down tweaks… a small improvement might require massive engineering effort.
    - rbanffy 3y ago
      Intel may be able to squeeze a little more juice out of x86 by dropping backwards compatibility. This is something I suggested they could have done for Apple because Macs were never supposed to boot MS-DOS anyway. A simple and regular ISA is what makes ARM easier to implement, which also means lower power because of fewer transistors doing thankless work like decoding instructions and reordering them.
pixl97 3y ago
Because the vendor will no longer give the the old chips? That an lack of support around security.
capitalsigma 3y ago
There are overheads in server workloads that scale with the number of machines (network traffic, serializing/deserializing requests). There are also fixed costs per server that don't scale with core count, or at least scale sublinearly (storage, physical data center space, motherboard, ease of maintenance). So running 10 machines with 100 cores can be cheaper and more performant than running 1,000 machines with 1 core even if $/core is higher. And of course individual cores can be beefier: wider SIMD units, application-specific extensions like bfloat support for ML workloads, etc. Of course Moore's law is slowing down, but cores/$ is an extremely silly metric to use

fsckboy 3y ago

The cores/money comparison doesn't looks like it accounts for inflation.* more to the point, the comparison also represents the period of time in which AMD, the one-time "lesser" player nipping at monopolist Intel's heels by competing on price, became the technology leader and started commanding a premium for their products. Meanwhile, Intel has not been forced to cede its position in the market, their existing contracts and business model is based on premium products, not cost leadership. It will take awhile for this to sort out.