Gigabyte GA-EP45-EXTREME Living Review

[Clunk]

5mm dense neoprene gasket not do the trick?

[eva2000]

i was thinking it would be also back if the spacing between backplate was too much as TRUE120's bolts are spring loaded and would require alot of force for those bolts to be pushed through the board mounting holes to reach the backplate ?

Anyway weird problems with my EP45-Extreme any modules installed in dimm 2+4 = instantaneous and continuous rebooting of the system ? Install in dimm 1+2 and not probs ?? Using F6a bios and 2x1GB Crucial Ballistix PC2-8500 D9GMH double sided modules.

[supershanks]

users have to be careful that they do not space it to much or else the backplate could ground to the case....of course eva, i know you would not let that happen
Edit/Delete Message

Don't think many have tried in a case yet
I will be monday I hope Gigabyte were in touch yesterday and the replacement board should be here then. Will conntinue review on water. I have a Scythe back plate I want to try , it's thinner than the thermalright, so will check the options.

[Alcibiades]

I installed a Thermalright Ultra 120 Extreme on the EP45-extreme and the EP45T-extreme using stock everything, I was not gentle either when mounting those spring loaded screws, but encountered no problems at all.
But I did have to dismantle the cosmos s's massive side panel fan so that the case could be closed. I even installed a Thermalright Ultra 120 Extreme straight onto my X38-DQ6 with the crazy cool backplate still attached and that was a slightly scary moment as I had to use a lot of force, that one was a real tight fit.
These boards are tough and very well built from my experience.

[supershanks]

I installed a Thermalright Ultra 120 Extreme on the EP45-extreme and the EP45T-extreme using stock everything, I was not gentle either when mounting those spring loaded screws, but encountered no problems at all.

Thanks for the update mate

But I did have to dismantle the cosmos s's massive side panel fan so that the case could be closed.

Oops
Remember seeing your post with the krazy cool still on , never did get round to trying that

[eva2000]

Man is it hard work to get this board running smoothly - so many auto power down/reboot and reset cycles!

Starting to tame the beast though



More photos here

Managed Super Pi 1M bench run @7x564FSB 4-4-4-15 but needed PL10 and 2.48v to do it ! Not fully stable as only had EP45 Extreme setup less than 48hrs ago

System:

• E8500 Q740A479T 2L7 / 1313
• CPU Cooling: Dtek Fuzion Rad Tower Box
• Gigabyte GA-EP45 Extreme rev 1.0 F6a bios
• 128MB Gainward FX5200 PCI
• 2x1GB Crucial Ballistix PC2-8500 dual channel kit
• Memory cooling: 120x25mm Thermaltake 81cfm fan
• 74GB Western Digital Raptor
• Pioneer 215 DVD-RW
• 1200W Silverstone OP1200
• WinXP Pro SP2 Nlite Fully Updated.

Super Pi 1M


Everest Bandwidth & Chipset
Seems that read memory bandwidth and latency are quite low for 564Mhz 4-4-4-15 memory clocks and timings but write and copy bandwidth look decent. Probably due to the loosened Static tRead (what DFI/Asus users would of known as tRD or performance level) of around 10. Gigabyte Easytune6 utility misreports, the values set in bios for CPU Reference, CPU Reference2 and MCH Reference. Easytune6 reports the default normal values instead they should be 0.853, 0.868 and 0.828 respectively.

[supershanks]

Originally Posted by eva2000

Man is it hard work to get this board running smoothly - so many auto power down/reboot and reset cycles!

found I started getting somewhere really when i adjusted GTLref & Ram sub voiltages Tuning Bios - GTLREF & DRAM Sub Voltages

Just got new board today from gigabyte , 1st one died , new one is the retail package like your's looks to be,

Getting Ram running seems to be the biggest challenge, though must admit haven't pushed it as hard as you have

[eva2000]

key to mem clocks seems to the same as other boards static tRead/tRD adjustments + vdimm + MCH volts

so far tested 5:6 2.0B and now testing 4:5 2.4A and same thing, static tRead might be set too tight by default, try loosening it by +1 or +2 and see if it opens up some memory clocking at decent reasonable vdimm

As you can see with loose Static tRead value of 10, Everest bandwidth and latency is slow and Super Pi 32M time is ~15 to 20 seconds slower than what is should be but not bad for low MCH voltage of 1.42v





Full report Gigabyte GA-EP45 Extreme rev1.0 Review - View Single Post

[supershanks]

static tRead might be set too tight by default, try loosening it by +1 or +2 and see if it opens up some memory clocking at decent reasonable vdimm

Yep this has always been finely tuned on the X38, X48 & now the Extreme, If exploring & using Auto tRD changing Performance Enhance from the default turbo to standard will use slacker tRD

I've had no problem using the red slots, there's not much difference but I find them slightly better.

I'm running at around 60.5%, 63.3% , 60.2% CPU Ref 1 , 2 & MCH Ref.
The bios options arte much clearer on this board, but that's pretty much where I think I was on the cruder X48T CPU GTLREF

Am just playing about with

Have bought myself the EP45T , not played much with that yet, struggling to find new bios.

Haven't got much done this week , had put my best E8500 aside to go in the ddr3 board , and set this up with my spare, couldn't hit the 8x533@2b 5-5-5-20 I had on the old board. to cut a long story short the spare E8500 looks like it's buggered , swapped over & it went 1st time.

Have an odd issue , though I have the most upto date Memset, most times I can't see Performance value

[supershanks]

Eva raised the matter of these chips interfering with backplates.
I didn't have the board at the time, so could only look at pictures. Have done a bit of research since :-

1) Thermalright Bolt Through
There is as George pointed out some interference. It isn't to noticeable but does give cause for concern


2) Scythe Universal Retention Kit
I'm lucky 'cos I had a couple of these kits

The backing plate on this very rigid steel , seems firmer than the Thermalright, but it also has an equal depth layer of neophane with a final layer of paper on top.


Side view showing the 2 layers.


I'm using 1 with my EP45 & my Fuzion block. A 2nd works well with the XIGMATEK HDT-S1283 on the EP45T.

This shot of the baseplate after unmounting, shows it's working pretty well
Gave a good impression anyway