Same memory, different day, worse score . . . ???

[brijem]

In January of this year (2012) I ran a full passmark performance test 7 benchmark;
beside a slight overclock on the CPU and the addition of
a GTX560 (replaced my 9600GT) the system is identical now
to what it was then, however, I recently (September) ran the same
benchmark (v.7) and the system scores a full 400 points less than it did
earlier in the year. The main drop is in memory performance - specifically the "Large Ram" mark.

In January my "Large Ram" score was 7820.
Now (with the same memory and no changes in the
bios concerning it) it has a "Large Ram" score of 1800.
Needless to say, it's a dramatic difference!
Also, the "Allocate Small Block" score is down by 40%.

Interestingly, my memory scores for "Read Cached", "Read Uncached"
and "Write" were all approximately 15% BETTER than they were
in January.

If it seems to be a factor, my CPU scores are also different.
They are down by a third in the "Integer Math" dept., down by
a sixth on "Floating Point", down by the same in "Multimedia"
and down by a sixth in "Prime Numbers". However, they are UP
by a small amount in "Compression", "Physics" and "String Sorting".

I'm confused (especially about the "Large Ram" drop).
Anyone have any thoughts?

Thanks

[David (PassMark)]

The large RAM test depends on how much free RAM you have available at the time of the test. This figure is unlikely to be the same as it was 9 months ago.

The other RAM tests do depend on the CPU performance to a fair degree in PerformanceTest V7. So overclocking your CPU could help the score in the RAM read tests. (In PT8 they are less dependant on the CPU and more on the RAM).

What type of CPU do you have?

[brijem]

I have a Phenom II X6 1100T OC to 3.7. The amount of free ram was about the same in the January test and the recent test (each time approx 13GB free). The other thing I meant to point out was that the recent system info (within performance test) lists my ram as 4 slots of 2GB - this is incorrect - it is 4 slots with 4GB in each slot . . . the January test lists it correctly. This is the only place where I've seen the RAM incorrectly reported - it's as it should be (4GB X 4) in the CMOS, etc.

[David (PassMark)]

The memory detection code was completely rewritten in PT8. So you might want to try that.

With the overclock there is the possibility that the CPU hits its temperature limit during some of the CPU tests and throttles back a bit. Higher clock speeds only mean better performance if the temps don't get too high.

Do you have the 2 baseline files available?

[brijem]

I don't think temps were much of a factor - I've run the test several times - a few times with all background processes minimized to only what matters and again with CoreTemp running at the same time as the benchmark, just to monitor - the cpu never got anywhere close to dangerous temps - 42c or so was the max (as I recall). Yes, I can post the baseline - I haven't saved the results of any recent tests and right now my computer is in the middle of a few things I cannot interrupt, but I'll retest it tomorrow and post the results, alongside the baseline from the January benchmark.

Also, I did try v.8 of the program and everything looks normal there - it sees the 4GB in each slot - only thing is that the "large ram" test does not seem to be part of v.8 (or it has another name) so I cannot compare that particular result with the v.7 result.

Thanks! I'll post those results soon. I appreciate the input : )

[David (PassMark)]

In PT8 the large RAM test is now just called the 'Available RAM' test.
As a general comment the PT8 and PT7 results aren't directly comparable, there are a few tests that haven't changed, but many have.

[brijem]

I'm thinking there's something wrong in my installation of version 7 - I ran another benchmark and suddenly it has decided to give me a new result - a total score of 836, down from 1482. Version 8 gives me a total of nearly 2200 and all the tests make sense and compare well to other similiar systems - version 7 originally had my system in the low 2000's, so I have to say that it's solved. I'm just going to start using v.8 from here on out. : )