DrediIt definitely is. Try measuring IPC on your PC without a motherboard, or main memory. I’m not holding my breath.

Vya DomusWhat a terribly unintelligent response. Was that supposed to be a gotcha moment or something ?

A processors executes instructions, not the motherboard, not the memory. IPC is a CPU metric.

defaultluserAnd Apple will become yet-another-Sun.

DrediBut the memory speed specifically has a high impact in how quickly the CPU manages to execute instructions, for some software.

Drediif you measure the IPC of some software with ddr5 and separately with ddr4 memory, which measurement is ”the IPC”, if the processor stays the same?

Vya DomusThen you're just measuring performance for that specific price of software, the CPU is still a constant that always behaves the same given the same input. If you however still decide to divide the number of instructions executed by the number of clock cycles elapsed while running that program and call that figure "the IPC", fine, but it's completely useless because it will vary for literally every single program for a million reasons.

Assume you run a program and you're trying to measure "the IPC", at some point that price of software has to communicate over the network, which obviously takes some time. At the end you draw the line and you say, well this CPU can do X instructions per clock. Do you see nothing stupid about that ? What did you actually measure ? The IPC of that specific CPU or system or whatever ? Really ?

Ultimately, if you want to measure IPC as precisely as possible you have to isolate the CPU from everything else, including memory speed. I don't know why that isn't obvious.



You tell me, according to you the IPC of a processor is an ever changing value dependent on everything and all values that you get are valid, somehow.

R0H1TWith 1000-10000x in lifetime profits ~ yeah not happening :twitch:

DrediBut the memory speed specifically has a high impact in how quickly the CPU manages to execute instructions, for some software.

if you measure the IPC of some software with ddr5 and separately with ddr4 memory, which measurement is ”the IPC”, if the processor stays the same?

And please, do explain how you measure the generic IPC you keep touting of.

TheoneandonlyMrKIPC instructions per clock is a hypothetical unbending max of the hardware. . .

TheoneandonlyMrKHow relevant is your version of IPC when everyone buys different ram SSD motherboard?! Then runs different games apps.

DrediJust describe to me how IPC should be measured, as you would do it.

ValantarI was responding to you stating that you don't think anyone buys $3000+ laptops for "office work", and your arguments against Apple knowing their audience. It's pretty clear that they do (in part because they've been pissing off their core creative audience for years, and are now finally delivering something they would want).

defaultluserI applaud Apple for pushing the boundaries, but their target audience for these machines is tiny - there will be no Linux or Windows support on these things!

The uptick in sales from the first generation ARM devics will die -off pretty quickly, And Apple will become yet-another-Sun.

Vya DomusSay you had two CPUs, one a lot faster than the other one but paired with memory that's a lot slower and you'd try and calculate IPC your way by running a memory intensive program and say that for arguments sake you find out both give you extremely similar figures. Then, according to you, it would be totally valid thing to claim that both CPUs have the same IPC, which would be obviously wrong and extremely idiotic.

Vya DomusI already said how, you try and remove as many variables that don't have to do with the CPU.

Vya DomusIt's absolutely important to know the lower bounds because that tells you what the worst case scenarios is.



Huh ? You always aim to use the most out of processor given whatever the time constraints are, I don't know what you are talking about.


That's such a bizarre thing to say. OK, you find out that a CPU can achieve X IPC in a certain application. What can you do with that information ? Absolutely nothing, people measure IPC to generalize what the performance characteristics are. If you are only interested in an application, as you say, then IPC measurements are pointless, you're actually just interested in the performance of that application.



I guess ? It's still stupid and that doesn't tell you anything about how relevant the hardware is if you argue that people who didn't actually needed one would buy it anyway.



I have no idea but I fail to see why it contradicts anything that I said. I just said that more L3 cache doesn't always translate to much improved performance. If your code and data mostly resides in L1 cache then messing around with the L3 cache wont do anything. Obviously real world workloads are a mixture of stuff that benefits more or less from different levels of caches or from none of them at all.


I read many of his articles and while they're very good I can't help but notice he has a particular affinity for everything Apple does.


A wide core with huge caches and most importantly a very conservative clock speed. That's why I am not impressed, trust me that if their chips ran at much higher clocks comparable to Intel's an AMD's chips while retaining the same characteristics then I'd be really impressed. But I know that's not possible, ultimately all they did is a simple clock for area/transistor budget tradeoff because that way efficiency increases more than linearly. I just cannot give them much credit when they outperform Intel and AMD in some metrics while using who knows, maybe several times more transistors per core ?

Darmok N JaladWhen M1 came out, porting for Linux began. They have already got it to booting to the GUI, though GPU acceleration is not there yet and is a bigger hurdle. Apple said running Windows on M1 is in MS’s court, but MS has shown no interest in licensing their ARM version of Windows, especially now that you can just license a hosted version of Windows. Parallels works on M1 Macs.

defaultluserI applaud Apple for pushing the boundaries, but their target audience for these machines is tiny - there will be no Linux or Windows support on these things!

The uptick in sales from the first generation ARM devics will die -off pretty quickly, And Apple will become yet-another-Sun.

DrediBut it would be the same!

defaultluserI'm sorry man - forgot that several million people ran Linux GPU-less on their PS3.

Oh wait, no they didn't :rolleyes:

If you have no GPU-support, then you have no public interest in the project (why would I spend gargantuan amounts of money for one of these things and shove Linux on it when you can't use the proprietary GPU or tensor system for anything?)

The PS3 worked better than Mac ARM Linux (because they had an official port,) but in the end-of-the-day nobody actually used it for home use! Mac ARM Linux, by-comparison going to remain a hack-job for years (therefoore, DOA!)

DrediNo it is not. You are mixing IPC with theoretical throughput (usually given as FLOPS/IOPS). What point is in your idea of IPC? What use is it to anyone? Can you link some text where it is used in that manner?



It is not. IPC is a metric one can use in the quest of understanding arcitectural differences, nothing more. It has little to do with actual real life performance.

TheoneandonlyMrKWho said it was useful to anyone.

It's in the abbreviation

Instructions per clock.

it's not my fault if you and others are using it wrong ,out of context etc.

Your on about application performance in an application.

I'm on about the actual term IPC.

It's been misused for years.

First note on wiki
"In computer architecture, instructions per cycle (IPC), commonly called instructions per clock is one aspect of a processor's performance: the average number of instructions executed for each clock cycle. It is the multiplicative inverse of cycles per instruction."

en.m.wikipedia.org/wiki/Instructions_per_cycle

I'm not at work now do go on.:)

And as a SYSTEM test engineer your reasoning is unsound you use known values and standards to ascertain or Benchmark the performance of a System not any or every application one might use.

You use design/mathematically provable and defined criteria to classify the spec of a Part.

TheoneandonlyMrKIPC instructions per clock is a hypothetical unbending max of the hardware. . .

DrediOh, you managed to read the wiki page finally. Too bad you didn’t understand the contents.

As you quoted, IPC is the average instructions per cycle of a given processor, (for a given piece of software). It has nothing to do with the maximum instructions per cycle of a given processor, what you previously wrote.

TheoneandonlyMrKIPC instructions per clock is a hypothetical unbending max of the hardware. . .

dragontamer5788That value is 6+ for Intel/AMD, depending on some details of uop caches I don't quite remember. If the uop cache is full, it drops to 4 (fastest the decoder can operate from L1). Furthermore, Zen, Zen2, and Zen3 all have the same value.

Which is nonsense. We can clearly see that "average IPC" in video game applications goes up from Zen -> Zen2 -> Zen3. IPC is... this vague wishy-washy term that people use to sound technical but is really extremely poorly defined. We want to calculate IPC so that we can calculate GHz between processors and come up with an idea of which processor is faster. But it turns out that reality isn't very kind to us, and that these CPUs are horribly, horribly complicated beasts.

No practical computer program sits in the uop cache of Skylake/Zen and reaches 6 IPC. None. Maybe micro-benchmark programs like SuperPi get close, but that's the kind of program you'd need to get anywhere close to the max-IPC on today's systems. Very, very few programs are written like SuperPi / HyperPi.