Anyone with a 3770K @ 5GHz on air?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Puppies04

Diamond Member
Apr 25, 2011
5,909
17
76
You're absolutely right, HT can decrease performance in IBT. I would expect an i7-3770K at 5.0 GHz to push close to 140 GFlops. A 20% reduction in performance as a result of HT being turned on is rather extreme - if my memory serves me correctly, HT usually only reduces performance by a few percent.

Not from what I have seen, my friend built a 2600K rig at the same time I build the rig in my sig. he was getting 90 something gflops (with HT turned on) while I was somewhere around 110 with us both at 4.5ghz.
 

VonDutch

Junior Member
Jan 10, 2013
10
0
0
I think your overclock may be a victim of phantom instability. The error checking on the CPU is correcting the instability, but your performance is being hit in the process of it.

My reasoning behind this is that my i5-2500K at 4.4 GHz pushes 120 GFlops. Your CPU appears to be slower, despite the fact that at 5.0 GHz, it should be 15-20% faster due to Ivy Bridge's minor architecture improvements.

this one i just did at 4.5ghz, no HT, 119 Gflops
LL


4.5ghz, with HT, 101 Gflops
LL

vcore i used was the same as with 4.7ghz..lol, only changed multi..im lazy ;)

i just talked to a member over at ocn,
he showed my one of his graph's, check the gflops, thats with HT on..
LL


i dont think gflops scales with ghz the same way you think it does,
and i dont think my chip is a "victim of phantom instability",
but i cant really check something that is not getting detected,
except if my oc's getting slower maybe?

o btw, hes one of the few running 5.0ghz 24/7 :D
 
Last edited:

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Gflops are also very sensitive to the specific problem size that gets ran, dependent on the ram used for the test.

You should get a dead-linear curve between clockspeed and Gflops.

GFlopsvsGHz.png


Running problem size 43122 with DDR3-1866 ram, Sandy Bridge will give you 28.28 GFlops/GHz and Ivy Bridge will give you 28.42 GFlops/GHz.
 
Aug 30, 2012
73
0
0
I'm getting 101 Gflops in IBT, though my volts seem to less than anyone else i've seen. Also no WHEA errors.
Are they too low? 1.235 in bios but LLC pulls it to 2.212 under load.
1.png
 

VonDutch

Junior Member
Jan 10, 2013
10
0
0
I'm getting 101 Gflops in IBT, though my volts seem to less than anyone else i've seen. Also no WHEA errors.
Are they too low? 1.235 in bios but LLC pulls it to 2.212 under load.
1.png

yea, thats about right (the gflops),
i need about the same vcore as you at 4.5ghz, 1.235V vcore :)

gflops arent very important,
when theres a big difference compared to others, then maybe,
its simple, if you start with 127 gflops at 4.5ghz like IDC,
you should end up at about 140 at 5.0ghz,
we start lower compared to him or pantsaregood,
but we scale the same way if we up our oc's..check the graph i showed from another member,
he goes up about 2 gflop for every 100mhz, give or take :)
same goes for IDC's graph
 
Last edited:

Vectronic

Senior member
Jan 9, 2013
489
0
0
101 seems a bit low, but I haven't used IBT in awhile... LinX with:

30250 problem size
7005 MB RAM usage (of 8GB)

LDA: 30264
124.8731 GFlops
147.795 seconds

3570K @ 4.5GHz
RAM: 2000 10/10/10/26/2

Don't have time to do a full 5 or 10 runs, so that's just the first run... usually the second or third is the fastest.

Edit:
1.235 in bios but LLC pulls it to 2.212 under load.
Whoa... what?... 1.235 to 2.212... ?... You meant 1.212?... RIGHT?...lol
 
Last edited:

VonDutch

Junior Member
Jan 10, 2013
10
0
0
Klunt Bumskrint and I have the same 1600mhz ram,
yours is 2000mhz, IDC used 1866mhz
you think theres a difference in that too when you run ibt?
 

Vectronic

Senior member
Jan 9, 2013
489
0
0
RAM speed/timings makes a difference, I don't think it would be 1GFlop for every 100MHz though.

Going by IDC's 28.42 * GHz (4.5) = 127.89... seems about right, I don't think I've ever hit ~128 at 4.5GHz though, 127 yeah, so it's close enough.

Edit: trying to use the same settings for both:

txPSNsN.png
 
Last edited:
Aug 30, 2012
73
0
0
Hijacking my own thread slighty.

But I had a go with the autotune on Gigabytes ET6.

It buggered my bios and I eventually had to reset it but it slowly bumps up the multiplier and tests for stability to find your max usable clock speed.

This is what mine was!
Captuhu.png

Temps didn't go above 71 either, not bad for 8.61GHz

Needless to say I've put everything back to 4.5 ( Reality )





Stupid Program!:rolleyes:
 

VonDutch

Junior Member
Jan 10, 2013
10
0
0
Hijacking my own thread slighty.

But I had a go with the autotune on Gigabytes ET6.

It buggered my bios and I eventually had to reset it but it slowly bumps up the multiplier and tests for stability to find your max usable clock speed.

This is what mine was!
Captuhu.png

Temps didn't go above 71 either, not bad for 8.61GHz

Needless to say I've put everything back to 4.5 ( Reality )


Stupid Program!:rolleyes:

sorry Klunt Bumskrin, my bad , shouldnt have asked about the so called "Phantom instability" in your thread :)

i think youre doing fine with your 8.61ghz oc ...LOL
forget about the 5ghz ;)

i tried ET6 when i first had my build together, but didnt work very well for me, still dont understand why it oc's using blck..my mobo doesnt work very well with higher blck's..
 
Aug 30, 2012
73
0
0
sorry Klunt Bumskrin, my bad , shouldnt have asked about the so called "Phantom instability" in your thread :)

Don't worry man!

I don't think you can beat doing things yourself. I've not seen even 1 program that can OC as well as good 'ol human trial and error.