CPU Simulation - Can You Simulate Performance of One CPU with Another?
Linus Tech Tips
·Linus Tech Tips
·2015-05-07
·
1,242 words · ~6 min read
0:00
in a recent video you guys called me out pretty hard it was the one here about
0:05
the performance of CPUs with varying numbers of processing cores when gaming
0:11
at the same time as performing other
0:14
intensive tasks in the background and many viewers were displeased with the
0:19
methodology that I used to obtain my numbers now of course I didn't dive
0:24
headlong into hours of testing without some kind of hairbrain theory about how
0:27
it would work so I was pretty sure my numbers were fine but one should never
0:32
be afraid to question one's own assumptions and whatever the result is
0:36
whether I was right or the critics were right I'm here to share them with you
0:41
just like soon I'll be sharing my thoughts on the upgradeability of the
0:44
iMac 27in retina 5K allinone PC that I've been using as my daily driver but
0:50
first the
0:57
intro ensure Swift Victory with Corsair
1:00
highly responsive lightweight gaming mouse the saber RGB click now to learn
1:04
more so I guess I should explain what people were so upset about before I get
1:09
too much further into this because the idea behind that video was to find an
1:16
extreme workload scenario first then see
1:19
how it affected a variety of processor configurations a fair amount of time was
1:24
actually spent developing The Benchmark scenario that I'd be using in order to
1:28
ensure that it put enough load on my
1:32
configuration to let the more powerful CPUs stretch their legs while not
1:36
putting so much load on my configuration that absolutely everything choked to the
1:41
point where it didn't function so I decided that instead of removing the
1:45
cooler swapping out the chip reapplying thermal compound then putting a new
1:49
motherboard on the test bench each and every time I needed to see how my stress
1:54
test was going to affect another CPU class I was going to Simply use the BIOS
2:00
of the motherboard that I was using to take my best processor an i7 5960x
2:05
extreme Edition and limit it artificially to the specs of the other
2:10
contenders in the test now hold on a minute lonus can you do that in the vast
2:15
majority of cases no no you can't I had
2:18
a lot of people ask why I didn't include AMD processors in that test and this is
2:23
actually the answer since the point of the video was more cores versus fewer
2:28
cores not Red Team Versus blue team anyway there was no real point and it
2:33
would be a lot more work because I'd have to swap benches and deal with
2:37
drivers in between runs every time all
2:40
right lonus so then in what situation would this artificial CPU faking
2:46
theoretically work if every CPU you're
2:49
simulating was designed with the same processor architecture so in this case
2:53
everything was Haswell based and has the same non-adjustable specs so cash memory
3:00
Technology support motherboard socket Etc and differs only in terms of the
3:05
parameters that we can change frequency multiplier hyperthreading support turbo
3:10
boost ratios Etc then they should
3:14
perform identically now I did Fudge things a little bit further than that
3:18
and simulated CPUs with completely different RAM support DDR3 versus ddr4
3:23
and different cash amounts but none of my workloads were hyper cash dependent
3:29
to my knowledge and I had already found DDR3 versus ddr4 performance to be
3:35
pretty much a non-factor outside of synthetic benchmarks in the Apples to
3:39
Apples comparison I had done previously which you can see here so I went for it
3:44
and here we are after all that Preamble ready to test whether my methodology was
3:49
okay or not cue the charts and graphs I
3:53
ran a variety of benchmarks on both a core i7
3:56
5960x that had memory channels cores hyperthreading and turbo boost adjusted
4:01
to simulate a core i5 4670k and a Pentium
4:07
g3258 and then I ran the same benchmarks on the actual chips themselves with the
4:12
only differences in theory being DDR 31866 RAM versus ddr4 2400 RAM and the
4:18
Oni cash sizes and honestly I was both
4:22
pleased with my theory because it's been
4:25
somewhat validated now cuz the results really are very close for the 46 70k and
4:31
also a little disappointed because I can see that as cash sizes really fall off a
4:36
cliff this Theory cannot be applied our
4:39
simulated g3258 outperformed our real
4:42
one by a fairly significant margin in some benchmarks so there you have it I
4:48
still stand behind the way I ran those tests because the time savings was
4:52
definitely worth the margin of error that was introduced by doing it that way
4:57
especially considering that the point of the video wasn't the exact models of the
5:00
CPUs but rather their number of cores and approximate multitasking proficiency
5:06
but I see that if I was looking for a way to save time on our normal CPU
5:10
performance reviews this obviously isn't it and we'll have to continue doing
5:15
board and Chip swaps moving forward especially if we want to measure things
5:19
that would be different no matter how close you get the specs due to Intel's
5:24
binning process like heat output and power consumption oh and speaking
5:29
speaking of Intel CPUs today's episode sponsor Gigabyte wanted us to show off
5:33
their new P35 V3 gaming notebook an
5:37
extremely slim gaming notebook which is powered by a fourth gen Intel Core A7
5:40
mobile quad core processor this baby features an NVIDIA GeForce GTX
5:46
980m a 15.6 in 1080P monitor the panel
5:50
type may vary by region it can support up to 16 gigs of DDR3 memory up to a 512
5:54
gig SSD and up to 2 terby of hard drive storage one of the other pretty amazing
5:58
things about this bad boy is that it's only 20.9 mm thick and weighs in at 2.2
6:03
kg it's extremely easy to transport and with its rear exhaust your hands should
6:08
remain cool and comfortable even during a heavy gaming session so if you're in
6:11
the market for a new gaming laptop oh yeah did I mention it actually has an optical drive they squeezed an optical
6:17
drive into this thing so sorry if you're in the market for a new gaming laptop
6:20
maybe you're getting you know way more Christmas money than I ever did or
6:23
perhaps you need to pick up a new notebook before the new school semester starts check out the Gigabyte P35 V3 get
6:29
the link in the video description if you want something that can run all the latest games as well so thank you guys
6:35
for watching this video like it if you liked it dislike it if you disliked it leave a comment letting me know if your
6:40
feelings were just more complicated than this as always don't forget to check out
6:43
the link in the video description to support us you can buy a cool t-shirt like this one you can give us a monthly
6:48
contribution or you can change your Amazon bookmark to one with our affiliate code so whenever you buy new
6:52
notebooks CPUs fans dry erase
6:56
markers or whatever else the case may be uh we got a small Kickback thanks again
7:00
for watching and as always don't forget to
7:20
subscribe