Look like I win a silicon lottery

[Sasqui]

Yes disable cores fore some more stable.

Ah, come on, that's not how we play the game!

 

[FreedomEclipse]

Yes disable cores fore some more stable.

sorry man, have to agree with sasqui

 

[TheHunter]

I see you have the same ram timings as me @ stock 2133mhz 9-11-10-27, did you raise VCCSA for 2800mhz CL12 1T or stock 0.808v?

Also can you please post AIda64 memory benchmark @ 2800mhz, really curious if you get full bandwidth, should be over 40-44gb/s for sure

 

[redeye]

how about prime95 for 24 hours?

 

[Vlada011]

I never saw such nice Haswell... But without 4 cores that's not real thing.
And I could imagine only temps with AVX instructions as Prime95 Blend, OCCT AVX Capable Linpack, All Cores 64bit... or LinX 0.6.4 AVX Edition or 0.6.5...
Probably 90-95 even on 4.5GHz...
Anything above 4.4GHz on Haswell is big success.

 

[Enterprise24]

I see you have the same ram timings as me @ stock 2133mhz 9-11-10-27, did you raise VCCSA for 2800mhz CL12 1T or stock 0.808v?

Also can you please post AIda64 memory benchmark @ 2800mhz, really curious if you get full bandwidth, should be over 40-44gb/s for sure

Nope VCCSA leave @ auto
Write is good but read and copy is a bit low , L1 cache read is insanely though 1220GB/S
@ 2933-3000Mhz bandwidth is huge drop I guess because too loose sub timing and it need huge voltage at 1.7-1.75V

 

[Enterprise24]

how about prime95 for 24 hours?

Didn't test for long time since temp are big issue in app that running with AVX2.
But I play CPU intensive game (like Civilization V and Total War Rome II) for 2-3days , It's rock solid

 

[Vario]

Didn't test for long time since temp are big issue in app that running with AVX2.
But I play CPU intensive game (like Civilization V and Total War Rome II) for 2-3days , It's rock solid

Rock solid is debatable if you can't run prime 95, prime 95 is the minimum for stability testing... LOL

 

[FreedomEclipse]

8hrs prime95 = minimum
12hrs to be more or less 100% sure if stable.
12hrs+ just for lols

Youre not fooling/cheating anyone but yourself when you disable 2 cores to get to 5ghz or not running prime95 or a least a long run of IBT before declaring your OC is stable.

as for your choice of 'stress testing' games.

Civ-V - "overall per-core usage levels out using ~50% of each core..." - Not really a 'CPU intensive' unless youre running a DC, TC CPU or locking the game down to 2-3 cores

Secondly I cant seem to find any CPU usage information on TW:R2 but i do have a CPU scaling chart

the game doesnt scale better with anything above a quad core. That being said, given the fact that the that 2 CPUs at the top of the chart are clocked at 4.8 & 4.9Ghz respectively Id go as far as saying that its the single threaded performance that counts most, from what i read TW:R2's engine is a complete mess and isnt really well optimised for multi-core CPUs

but it looks like you got a better answer from crazyeyesreaper

It might be CPU intensive but its no good when its only locked to 1 or 2 cores.


Like i said - Youre fooling no one here but yourself.

 

[Sasqui]

Rock solid is debatable if you can't run prime 95, prime 95 is the minimum for stability testing... LOL

I'm agreeing with both Vario and "Crazy Dogmatic Bullshit"! I wouldn't consider a rig stable unless it ran Prime95 on all cores for 12 hours straight. I got that on 4.8 Ghz and backed it down to 4.7 Ghz for a 0.1 Ghz safety factor

 

[TheHunter]

Nope VCCSA leave @ auto
Write is good but read and copy is a bit low , L1 cache read is insanely though 1220GB/S
@ 2933-3000Mhz bandwidth is huge drop I guess because too loose sub timing and it need huge voltage at 1.7-1.75V

Ah ye that looks bugged a bit, mobo sets 2nd timings too lose.


Imo you would have better luck with 2400mhz cl10-12-12-31-1T. I bet 35.7 gb/s by read and copy

 

[Enterprise24]

For Prime95 long time test it must be delidded.

Some test for 4.9Ghz

Test System Setup
i5-4670K @ 4.9Ghz 1.3V , Uncore 4.7Ghz 1.305V , VCCIN 1.9V
Cooler Master Seidon 240M
Gigabyte Z87X-OC BIOS F7d
Avexir Core Series 2133Mhz CL9 @ 2800Mhz 12-14-14-35-1T 1.65V
Plextor M5Pro 256GB Firmware 1.06
XFX R9-290 @ R9-290X Default Clock (1000Mhz/1250Mhz) Catalyst 14.2 beta
Seasonic X Series 750W
Windows 8.1 Pro x64

2D

















3D









3Dmark 11 and Fire Strike seem to have driver issue.

 

[Enterprise24]

Delidded with vice method and apply liquid pro temp drop around 20C.

Now 5Ghz with Aida 64 stability Test is possible

 

[Enterprise24]

33minutes should guarantee 90% stability

 

[Vario]

are all 4 cores enabled, I see 4 cores 4 threads in cpu-z just curious

 

[Enterprise24]

yes all 4 cores enabled

 

[Vario]

Very nice!

 

[Enterprise24]

Try Intel XTU for 1hr
Still love this chip , I think it's better than devil's canyon.

 

[Ed_1]

What app is that your adjusting memory timings in windows ?

AsRock Timing Configurator ?
but your using Gigabyte MB

 

[OneMoar]

insane amounts of voltage I am surprised it didn't smoke when you ran 1.5V though it
5Ghz @ 1.5 the FIVR has gotta be smoking hot did you put thermal tape over the FIVR/cache on the top side of the chip so it makes contact with the WB ?

 

[Enterprise24]

What app is that your adjusting memory timings in windows ?

AsRock Timing Configurator ?
but your using Gigabyte MB

Nope , I change MB from Gigabyte z87x-OC to Asrock Z87 OC Formula.
I think Haswell cannot change timing in windows but Ivy Bridge can.

 

[Enterprise24]

insane amounts of voltage I am surprised it didn't smoke when you ran 1.5V though it
5Ghz @ 1.5 the FIVR has gotta be smoking hot did you put thermal tape over the FIVR/cache on the top side of the chip so it makes contact with the WB ?

1.4 not 1.5V
Kraken X60 contact IHS not die.

 

[OneMoar]

1.4 not 1.5V
Kraken X60 contact IHS not die.

thats why I asked if you added a strip of non-conductive thermal tape to fill the gap between the block and the die

 

[redeye]

For Prime95 long time test it must be delidded.

Some test for 4.9Ghz

Test System Setup
i5-4670K @ 4.9Ghz 1.3V , Uncore 4.7Ghz 1.305V , VCCIN 1.9V
Cooler Master Seidon 240M
Gigabyte Z87X-OC BIOS F7d
Avexir Core Series 2133Mhz CL9 @ 2800Mhz 12-14-14-35-1T 1.65V
Plextor M5Pro 256GB Firmware 1.06
XFX R9-290 @ R9-290X Default Clock (1000Mhz/1250Mhz) Catalyst 14.2 beta
Seasonic X Series 750W
Windows 8.1 Pro x64

OMG... you are using the F7 BIOS on a z87!... don't you find it slow?, compared to an msi or asus?


i do, i upgraded my h87n-wifi from F4 to F7 so i could try a i5-4690... BIG MISTAKE the F7 BIOS reduces my 3dmark fire strike graphics test FPS by about 4.5% to 3.6 percent!...
Do not upgrade h87 motherboards!... never the h87n-wifi. (until F8 fixes the slow-down)

in anycase my F7 gigabyte h87n-wifi, is slower than my MSI h81!!!... or asus h87i-plus (and i threw caution into the wind, and upgraded the Bios on the asus to 1008... No problems... same speed)

so, i am angry that gigabyte can f*ck-up a Bios update... That i will never recommed a Gigabyte motherboard unless you are using it for a customac/hackintosh!...

really?, an msi h81i with a pcie2.0 slot is faster than a gigabyte h87n-wifi F7 pcie 3.0 slot (both using a i3-4330)

http://www.3dmark.com/compare/fs/2227786/fs/2212916

http://www.3dmark.com/compare/fs/2229889/fs/2229706

http://www.3dmark.com/compare/fs/2269030/fs/2253589

(yes, it is possible that Gigabyte did not f*ck-up the F7 Bios on a z87, because they eliminated all of overclocking stuff on the h87 f7 bios, but you might want to check against another manufacturer's board... i only realized this when crysis3 was not running as smooth as it was on F4 bios.)

 

[Enterprise24]

F7 bios has issue with sub timing , I can't change some of it. So I sell it and get Asrock MOBO.