p67 ext 6,clocks and gen 3 bios + memory speed

[kick]

When I originally bought my system back in early 2011:
I was doing 4.8 ghz with 1333 memory
cpu v was about 1.47
default v for the gskill ripjaws was 1.59v
Now with lots of testing recently.Ive settled on

4.6 cpu speed(1.42-1.43v max) standard overclock but with llc at 3 in bios
gskill mem (sold as 1333 mhz 9-9-9-27).. and works as such at 1.5-1.59v)
now at 10-10-10-30 1600mhz and 1.62v

Machine is stable under prime for 10 hrs with some benchmark heaven testing every so often to warm up overclocked 460's.Kinda jerky with prime running at the same time but np..

Memory at 1.59v auto..it wont do 1600mhz
Cpu gets to 68-71c.
I had been considering some extra memory .. maybe getting some 8g gskill ripjaw certified at 1600 stuff .. but even at $40 the performance difference would be unnoticeable.

Its weird: a 2 year old pc and there isnt any real difference between what I have and a 90% brand new model in the real world.
Maybe next cmas if the ps4 hasnt killed off the pc :)

[parsec]

You lost me here, what is the point? No difference between 1333 and 1600 memory? That's true, except in benchmarks as you implied. Usually very little return on using memory above 1600 too. Memory speed is not the bottleneck in any PC, as long as it's at 1333 or more.

[kick]

:) Its like this.

My second system listed, circa 2008, I could obtain a noticeable fsb clocking increase for both memory and cpu.
Now 2133mhz memory similar to my 1333mhz memory is available for about the same price(system1), but its hardly worth the effort of fitting it,paying for it.

I can run the i5 happy at 4.8,very happy at 4.6and so on but the 1333 mhz memory doesnt make a lot of diff.I could pull it out and put in some 2133 capable items ,but why bother,
The 20008 system I had ,even clocked up is real life a bit worse than my 2011 system but aside from benchmarking stats , there isnt any real incentive to change my early 2011 system .games ,net, speed of operation is still ok after 2 years. Thats unusual in pc land.

[Conners]

I read Kick that anything above 1600MHz memory is a waste of time. Although I am running my samsung 1600 @ 2133MHz. Does it make a diff? Well does it?

[profJim]

Kick, 7 months ago I asked Does DDR3 Memory Speed Really Matter with Ivy Bridge or Sandy Bridge?
Knowing what I know now about ssd raid 0 performance with high end enthusiast ssds, an interesting real world test might be to test with intensive applications such as rendering, sophisticated photo manipulation, or audio/video conversion when using a raid 0 ssd setup to see if much higher memory speeds really do improve performance. No hard drives would be allowed.

Several z77 threads in the Gigabyte forums reported somewhat lower benchmark scores with ultra high memory speeds, even with extensive tweaking.
When I upgrade to Haswell next year, I'll probably buy 32GB of1600MHz memory and use my "patented" 492MHz settings.
Maybe I'll use a 24GB ram disk to really make things fly.

[parsec]

I see, yes, changing to Ivy Bridge will buy you nothing performance-wise, given your apparent usage. There are differences in video conversion and rendering between your i5-2500K, and an i5-3570K, the latter being faster.

The major change that occurred in Intel CPUs, from Nehalem and earlier Core CPUs, like your Q8400, with BCLK and FSB OCing, that affected the CPU and memory, to the locked down BCLK in SB, and no direct access to the memory multiplier, and the CPU multiplier only OC of the CPU, is part of the situation. Personally, do I find my X58/i7-930 CPU with triple channel 1600 memory, in the same league as my P67 or Z77/i7-2600K or i5-3570K CPU with dual channel 1600 memory? No, the SB PC is faster. The IB system is perhaps a tad faster, partially due to the crazy Samsung memory, at 2133, 10 10 10 24 1T.

Part of that is due to a good SATA 6Gb/s interface on the SB system, and the use of SSDs. As much as I've tried to get the i7-930 PC to be as quick as the others, the SATA II interface makes a difference, even with good SSDs in RAID 0 volumes.

Haswell will probably not bring that much more CPU performance to the table than IB, but the addition of more SATA 6Gb/s ports is the major attraction for me. Forget benchmarks, when a "quick" virus scan on the OS volume can finish in under 10 seconds with a single or RAID 0 of SSDs, vs 10 minutes+ on a HDD, I too am not worried about CPU and memory performance, I/O is the final bottleneck, which still has not caught up. You have one SSD now, that is where I suggest you invest your PC funds.

[kick]

Oh hey heres my latest isopropyl alcohol fan cleanup..dust heat!




Hey Pro J, P sec and all, sorry C I cant respond pm but generally no it doesnt help unless you can maintain the 1600 clockspeeds on the memory. :(

As for spending cash on items for cmas :0. thats the problem .. terrible isnt it !.
ssd 3 and 4 altho 120gb 3 is available for around the same price as my 60gb 2 now,its not worth buying for my usage 2 years in.
The same goes for practically everything else.
There isnt the noticeable performance upgrade available,
Sure I could work to get my cpu cool and stable at 5 ghz with the latest bios approximated memory speeds ,but there isnt anything in a "quantum leap" vein out there.
Good news:
Perhaps ,but there is nothing Im so far behind the curve on that its worth spending cash(an exception is a new psu fror my 775 system,posted elsewhere.)
Time to say goodbye to my 480W(haha) Advance psu..yes ProJ the one that blows up as we discussed about 2 years ago.its still working but is tempremental.
The unit is still for sale btw altho now double my $17 original purchase price!.

One other thing I would buy if I could find a working model is a combined 5 1/4 in bay 5 1/4 in 3 1/5 in floppy drive for the cool retro look since the asrock has a fdd connector and :0


EDIT:Something I forgot to mention, for those interested:
The memory I have from new is the 9-9-9-24 1.5v 1333mhz gskill stuff.
This auto's on the mbrd to those timings at 1.59v.
In practice I allways leave it at 8-8-8-22 and the 1.59v ,for fsb 99.8-100.2 mhz cpu 46,47,48 never any prob. 7-7-7-20 sub 1.6v only ok on a very mild clock cpu
5 ghz cpu reliable needs slow timing n all that..not worth it.
and like that.
Memory readout:
CAS Latency (tCL) : 6 clocks @ 457MHz, 7 clocks @ 533MHz, 8 clocks @ 610MHz, 9 clocks @ 686MHz
RAS to CAS (tRCD) : 6 clocks @ 457MHz, 7 clocks @ 533MHz, 8 clocks @ 610MHz, 9 clocks @ 686MHz
RAS Precharge (tRP) : 6 clocks @ 457MHz, 7 clocks @ 533MHz, 8 clocks @ 610MHz, 9 clocks @ 686MHz
Cycle Time (tRAS) : 17 clocks @ 457MHz, 20 clocks @ 533MHz, 22 clocks @ 610MHz, 25 clocks @ 686MHz
Min TRC : 23 clocks @ 457MHz, 27 clocks @ 533MHz, 31 clocks @ 610MHz, 34 clocks @ 686MHz

[parsec]

OTOH, I fired up my old Gigabyte P43 board, socket 775 with an E8400 CPU, memory at 1066, and Vista. Even with a SATA II SSD for the OS, I was surprised how relatively slow it is compared to my other PCs. What surprised me was the high amount of CPU usage, when installing Windows Updates. Sure, two cores, no HT, so over 50% CPU usage. That is in comparison to single digit CPU usage for an i5-3570K.

So the older systems clearly are limited in CPU resources compared to new ones, and when multiple user processes, as well as other potential system processes are running together, the older systems performance will be worse. But, upgrading from your SB PC, yes, will not buy you much performance.