8 Packs 4ghz 32m Challenge win full Haswell 4770K Bundle.

Greebo said:
It used to make a massive difference defragging your hard drive prior to a test run but not sure that makes a difference now with ssd drives?
Click to expand...

Nah I don't believe it makes a difference.

Ran test on my f1 with less than 10% space remaining, sure there is a lot a thrashing noises, but moving over to my m4 made no improvement in times.
 
Gzero said:
Nah I don't believe it makes a difference.

Ran test on my f1 with less than 10% space remaining, sure there is a lot a thrashing noises, but moving over to my m4 made no improvement in times.
Click to expand...

It probably won't now since a multithreaded CPU will absorb the IO load. However consider the good old days when you had a single core to work with, your cpu will have to share processing PI along with disk IO.

I noticed a fair bit of disk activity from pi today and can see how this would have been an issue in years gone by.
 
Gzero said:
Nah I don't believe it makes a difference.

Ran test on my f1 with less than 10% space remaining, sure there is a lot a thrashing noises, but moving over to my m4 made no improvement in times.
Click to expand...

Surprised. in the old days with i7 and q9650, a none defragged drive made a difference of a few seconds. We always used to defrag before trying to set records. Thought going from a normal drive to a ssd drive would make a difference.
 
You could create a ram drive and run it from that, its possible that might improve times a little.

There's plenty free ram drive software to try.
 
Yeah used to try ram drives in the old days as well. Can't remember if that made a difference or not. Copy waza definitely gives you about 3 seconds though.

Doing a couple of 16k runs before the main event always did though as well.
 
There's a drop off point with dram,no matter how much you chuck at it it simply won't boot unless you can drastically cool them,1.7v is the highest I've gone,mostly 1.65v is plenty
 
I think you need Chinese brand 1066MHz with slack timings. Single channel. :D
 
Last edited:
Duff-Man said:
:p Sorry...

I'm sure I won't be the only Haswell entry, if it's any consolation. There's another 4 weeks to go yet.
Click to expand...

Lol what are you apologising for? :p

If I had a Haswell based system and I had the funds to buy some benching memory I would go for the top prize as well. :)

I just don't agree with the structure (or rather lack of one) of the competition :D
 
Duff-Man said:
Okay - very last question on memory from me - I swear :p As Jose Mourinho said - you need the right eggs to make a good omelette :p

2400Mhz 9-11-11-28

vs

2600Mhz 10-12-12-31

Leaning towards the 2.4Ghz CAS9 set, given Greebo's quantification of the effect of CL and tREF.
Click to expand...

hmmm. the 2600 is 8% faster but should around 1% slower for the slacker timings so the 2600 will be faster.

However, it is likely (opr used to be with memory) that you can run the 2400 memory at 2600 with better timings than the 2600 set
 
wazza300 said:
There's a drop off point with dram,no matter how much you chuck at it it simply won't boot unless you can drastically cool them,1.7v is the highest I've gone,mostly 1.65v is plenty
Click to expand...

Yeah which is why serious people watercool their ram whcih I have done in the past ;)
 
acme15 said:
Lol what are you apologising for? :p

If I had a Haswell based system and I had the funds to buy some benching memory I would go for the top prize as well. :)

I just don't agree with the structure (or rather lack of one) of the competition :D
Click to expand...

Well I agree with you, and I admit I would feel kind of bad about it if did win, especially if I were the only Haswell entry. Ideally the competition would have been run pre-Haswell release to prevent this kind of "sandbagging", but nevermind.

In my defense I'm coming from an i7-930 that I've had for over three years, and I was going to upgrade regardless of this competition. Perhaps I would have spent £20 or £30 less on the memory but meh - it'll give me an end goal for my tinkering :) And if I do win then I won't be selling the kit or anything - it'll be used for running large-scale simulations of underground CO2 sequestration using PFLOTRAN with my company. Duff-man -> solving global warming one CPU clock at a time :p
 
Greebo said:
hmmm. the 2600 is 8% faster but should around 1% slower for the slacker timings so the 2600 will be faster.

However, it is likely (opr used to be with memory) that you can run the 2400 memory at 2600 with better timings than the 2600 set
Click to expand...

Thanks...

So, all things being equal, you would go for the 2400Mhz C9 kit?
 
Duff-Man said:
Thanks...

So, all things being equal, you would go for the 2400Mhz C9 kit?
Click to expand...

I would assuming that it can run at 2600 with better timings than the 2600 memory but no gaurantees.

I only know about the ram around when I built my i7-950 rig and bought good ram with tight timings which were known overclockers.
 
You must log in or register to reply here.
Back
Top Bottom