Just wondering why my 8 core XTU sub not getting points?

http://hwbot.org/submission/2422247_dhenzjhen_xtu_xeon_e5_2687w_v2_1582

 

 

Thanks,

DJ

Global Points (UGP) 1st using 8 CPU - 11.0 Points

Hardware Points (UHP) 1st using Xeon E5 2687W V2 - 2.0 Points

Global Team Power Points (GTPP) 1st in team using 8 CPU - 50.4 Points

Hardware Team Power Points (HTPP) 1st in team using Xeon E5 2687W V2 - 5.2 Points

 

Seems fine here?

Hmmm that's odd because I saw while ago only UHP and HTTP

 

Hardware Points (UHP) 1st using Xeon E5 2687W V2 - 2.0 Points

Hardware Team Power Points (HTPP) 1st in team using Xeon E5 2687W V2 - 5.2 Points

That benchmark is flawed. I got the same score as you using the defaults and OC'd to 3871MHz..36x107.55BCLK

^ Don't run CINE more than 2 times in a single session, always restart as score starts dropping. BTW, if you're changing BCLK in Windows do it with Cine closed as it internally defines how long 1 second is using QPC.

^ Don't run CINE more than 2 times in a single session, always restart as score starts dropping. BTW, if you're changing BCLK in Windows do it with Cine closed as it internally defines how long 1 second is using QPC.

 

Hi, if you meant me all my changes are done in BIOS, no way to change in windows on these dual socket machines that I am aware of..I wish there was!

BTW, if you're changing BCLK in Windows do it with Cine closed as it internally defines how long 1 second is using QPC.

Isn't this a benchmark security leak? or it doesn't work the other way around?

It's a problem with every application using CLOCKS_PER_SEC (which is calculated according to CPU MHz) instead of the actual number, 1000000.

and yet it has both hardware and global points enabled?

If we want 100% secure benchmarks only I think we can remove globals altogether. Even the wrappers are not safe (though safer than no wrapper at all!).

 

I wonder if superpi/pifast are safe. My guess is that they're very easy to manipulate, especially pifast.

I wonder if superpi/pifast are safe. My guess is that they're very easy to manipulate, especially pifast.

May be I'm missing something, but to cheat SuperPi or PiFast by any margin, you would have to deliberately use "knowledge" which isn't normally used for overclocking/tweaking and hence deemed illegal ... whereas with CineBench you can be speedhacking on others without even knowing about it.

That benchmark is flawed. I got the same score as you using the defaults and OC'd to 3871MHz..36x107.55BCLK

 

Using v4.2.0.8 ?

Using v4.2.0.8 ?

4.1.0.12 Just downloaded it 2-3 days ago