A about the 7-zip Bechmark. It clearly focus on the MIPS standard..
After bying a new CPU I experienced following...
Everyone surely likes to know how much faster CPU they get for the money. I looked at the CPU passmark site on http://www.cpubenchmark.net and found that a...
Intel I3-3240 dual-core 3.4 Ghz
was about twice as fast as my old
Intel E8400 dual-core 3.0 GHz
But after testing the I3 CPU, I was very dissapointed to experience that the old E8400 was up to 24% faster than the new I3, according to the 7-zip MIPS benchmark (the I3 setup as dual-core only to compare equally).
Firstly I would blame the passmark test. If You look at the test which includes al lthe cores You get a wrong picture. The E8400 has 2 cores and the I3 allso, but with 2 virtual extra cores.
Comparing 2 cores with 4 cores gives a totally wrong picture because 4 cores would inadvertly be faster in any case (faster is a wrong term, more in the pocket is more correct).
That said, the passmark allso have a single core test graps and this shows that the I3-3240 has a mark on 1.8 against the E8400 with a mark on 1.25 (the I3 should be 44% faster).
And guess what: the 7-bench show that the E8400 is still faster than the I3 in single core mode with a MIPS of 3730 against 3191 on the I3 (The E8400 is 17% faster).
So is there any explanation ?
Intel i3-3240 must be faster than Intel E8400 in 7-Zip LZMA benchmark, if both CPUs are not overclocked.
Show all result numbers that you have, including compression / decompression rating numbers.
Write also about CPU frequency.
Check that you use same version of 64-bit 7-Zip.
I tried to make the test as equal as possible. I disabled the virtual cores on the I3-3240 but I did not change the standard clocking of the CPU's aka 3.0 and 3.4 GHz.
The 7-zip ver. is 9.20 32bit
I have attached dump from tests of both. From those i calculated this
I3-3240 : 5.420 MIPS / 3,4 GHz = 1.594 Mips/GHz
E8400 : 5.933 MIPS / 3,0 GHz = 1.977 Mips/GHz
Result : (1.977 - 1.594) / 1.594 = 24% faster (the E8400)
Probably you have disabled one core in i3. So 2 threads worked on one core.
You must enable both cores in BIOS and you must disable hyper-threading feature in BIOS.
hehe - ok - I allso enable all bugs in 7-zip menu -> features -> bugs
I'll be back
Well that helped.
Who knows that if You only enable one core on the I3, the beast wil still use one hyper-thread and show 2 running cores. I disable the hyper-dudes because it's more to run run viruses and other NSA spying in tha back ;). (Just gimme tha die-hard cores and leave tha cheap vudka to the noobs).
So now I can come up with these results:
I3-3240 (new)..: 7.339 MIPS / 3,4 GHz = 2.158 Mips/GHz
E8400 ...........: 5.933 MIPS / 3,0 GHz = 1.977 Mips/GHz
The I3-3240 advantage is then (2.158 - 1.977) / 1.977 = 9,1% faster
note that i normalize the MIPS by dividing by the clock so we can compare MIPS pr. GHz.
So getting a I3-3240 (3,4 GHz) in favor over an older E8400 is not worth the money as You would'nt notice the difference in speed.
However, You get the Hyperthreading which may be advantage in cases it is utilized (not for MS office or mail or blogging) AND some lower power consumption.
Thanks Igor for Your attention and sorry for the mishaps. 7-Zip is a great program.
Note also that
1) i3 with enabled hyper-threading is much faster than e8400 for 7-Zip.
2) Intel E8x00 cpus have largest L2 cache (6 MB) from all CPUs at long time. For example, i3 core uses 256 KB of L2 cache and 3 MB L3 cache.
So Intel E8x00 can be faster for some things than any other Intel cpu. The only problem of E8x00 is slow RAM via chipset.
Yo - Igor !
There's a little thing in the Bench. You can select more than actual cores in the "Number of CPU's" drop-down.
In my case I can select 6 cores despite the I3-3240 have 2(+2) and the Bench shows /4 (as max). Nevertheless I found that the test shows the same results as 4/4 was selected (I thought You had found a way to enable 2 extra cores or even invented a new super core extension. ok - another day).
You can read about benchmark threads here:
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.