What the fark is an “emulated” benchmark?

  • Thread starter Thread starter User.191
  • Start date Start date
U

User.191

Guest
@Cmaier you’re a man who apparently can spell CPU - so tell me: What the fark is this joker whittering on about?

forums.macrumors.com

Emulated 12900K for mobile has better performance with low power consumption like M1 Max?

https://meeco.kr/ITplus/33874440 Just an emulated result but according to China, 12900K with 6 big cores(3.0ghz) and 8 small cores(2.4ghz) used only 35W of power usage and the Cinebench result was around 14300 which is similar to M1 Max's result. If this is true, then Intel is able to make...
forums.macrumors.com forums.macrumors.com
 
MissNomer said:
@Cmaier you’re a man who apparently can spell CPU - so tell me: What the fark is this joker whittering on about?

forums.macrumors.com

Emulated 12900K for mobile has better performance with low power consumption like M1 Max?

https://meeco.kr/ITplus/33874440 Just an emulated result but according to China, 12900K with 6 big cores(3.0ghz) and 8 small cores(2.4ghz) used only 35W of power usage and the Cinebench result was around 14300 which is similar to M1 Max's result. If this is true, then Intel is able to make...
forums.macrumors.com forums.macrumors.com
Click to expand...

I don’t know. It seems to be gibberish. Not sure what the results are supposed to mean.
 
Cmaier said:
I don’t know. It seems to be gibberish. Not sure what the results are supposed to mean.
Click to expand...
The OP there seems to be very sure they mean great things.

I’m no chip gal, but even I was having a hard time trying to understand how you get “emulated” benchmarks.

Guess the AL fanboi is doing their best to try and stake a claim that somehow the power hungry chip ain’t all that power hungry - as long as you close your eyes tightly and squint somewhere else…
 
MissNomer said:
The OP there seems to be very sure they mean great things.

I’m no chip gal, but even I was having a hard time trying to understand how you get “emulated” benchmarks.

Guess the AL fanboi is doing their best to try and stake a claim that somehow the power hungry chip ain’t all that power hungry - as long as you close your eyes tightly and squint somewhere else…
Click to expand...
Sounds like somebody turned off a couple cores in an actual chip, misunderstood the power numbers (confusing processor power with wall power), and misinterpreted some conclusions.
 
No, why, I think it's fair to emulate a configuration of a particular SKU like this, assuming that they set up things correctly. Of course, the results are misleading, especially when used to drawing comparisons to M1. Not least because the benchmark (Cibenech) is not representative at all of M1's performance.

But I think in the context of comparing this chip to AMD Cezanne — mobile ADL might shape up to be a winner.
 
leman said:
No, why, I think it's fair to emulate a configuration of a particular SKU like this, assuming that they set up things correctly. Of course, the results are misleading, especially when used to drawing comparisons to M1. Not least because the benchmark (Cibenech) is not representative at all of M1's performance.

But I think in the context of comparing this chip to AMD Cezanne — mobile ADL might shape up to be a winner.
Click to expand...

Maybe. Hard to know how things fare when the ratio of P-to-E cores changes.
 
MissNomer said:
The OP there seems to be very sure they mean great things.

I’m no chip gal, but even I was having a hard time trying to understand how you get “emulated” benchmarks.

Guess the AL fanboi is doing their best to try and stake a claim that somehow the power hungry chip ain’t all that power hungry - as long as you close your eyes tightly and squint somewhere else…
Click to expand...

Colloquialism I guess. At first I thought it meant some kind of throughput estimation generated through software emulation. Intel publishes enough stuff to do this, since their architectural optimization manuals cover things like what ports may execute what instructions and some details on store forwarding, macro fusion, cache associativity, and some TLB details. Between that, the intel intrinsics guide, and Agner Fog's latency tables, you can put together decent expected latency, noting that Intel is overly optimistic when it comes to estimating load and store throughput. You could probably get an okay estimate by emulating this in software, particularly for workflows that tend to be somewhat oblivious to exact cache size. I haven't tried it myself, so I too might be overly optimistic. It's hard to say.

Edit: I meant to say, the person on reddit who claims to know "Chinese" (presumably Mandarin) gave a plausible explanation.




leman said:
No, why, I think it's fair to emulate a configuration of a particular SKU like this, assuming that they set up things correctly. Of course, the results are misleading, especially when used to drawing comparisons to M1. Not least because the benchmark (Cibenech) is not representative at all of M1's performance.

But I think in the context of comparing this chip to AMD Cezanne — mobile ADL might shape up to be a winner.
Click to expand...

Not related to the thread, but welcome aboard.
 
Back
Top