noob2222 :
juanrga :
noob2222 :
I find it funny when people cant even read an article properly. Gamerk hit the nail on the head, ipc is a theoretical "maximum" number thats affected by multiple aspects during any given test program.
No. IPC can be theoretical or can be measured; it can be maximum, minimum, or average. It depends of what you are doing and how. In the Extremetech article the IPC they obtained is an average value obtained from measuring the IPC of different benchmarks and taking the average.
benchmarks measure program specific performance not IPC.
IPC can be measured and is routinely measured.
noob2222 :
Extremetech article never claimed they were showing IPC. Go to the article and search for "IPC" and tell me how many times it shows up.
Moot point. The article is about IPC, but they named it differently. They used the term "efficiency", which is odd because efficiency usually refers to performance per watt.
noob2222 :
Instructions Per Clock is primarily a meaningless term in today's computing world, IPC ≠ performance.
As explained before by myself and others
Performance = IPC * frequency
IPC gives a measure of the kind of architecture. Extremetech is using it to discuss why AMD would abandon big cores and focus on improving Jaguar. I agree with Joel.
noob2222 :
you can't improve IPC by improving the performance of a program, your merely "improving the performance of said program", your not improving the performance of someone's product. you didn't go into the hardware itself and rearrange the transistors to make your program run faster, you didn't change the IPC. Instructions Per Clock are hard wired to a specific CPU. Instructions Per Clock isn't variable.
As explained in the link that I gave before but you insist on ignoring:
The number of instructions executed per clock is not a constant for a given processor; it depends on how the particular software being run interacts with the processor
http://en.wikipedia.org/wiki/Instructions_per_cycle
A compiler can increase the IPC not only by eliminating useless instructions but by adapting/rearranging the final code to the specific characteristics of a given processor architecture. This is why we have flags specifics for a given architecture (e.g., "Bdver3" for Steamroller). When the flag is selected the compiler generates a binary optimized for the Steamroller architecture and in general this binary will run faster than a binary is not optimized. For instance with the generic K8 flag C-Ray spends 87.90 seconds on Kaveri, when selecting the more specific Barcelona flag the execution time reduces to 53.33 seconds, and when selecting the Bdver3 flag specific to Steamroller it further reduces to 40.54 seconds.
Since the frequency of the processor was not changed the reason why the more optimized binaries runs faster is because the IPC has increased.
There and thereafter I will ignore your further posts about IPC.
Facebook
Twitter
Instagram