speed.fastest

Looks like Steamroller had better memory performance than kaveri at stock. Tweaked kaveri in AIDA i can get around 25K read @ 4.7ghz with around 2500mhz single rank memory.

So try monitoring VRM temp and Core Clock when benchmark running if they throttling or not? You can use Afterburner OSD Monitoring.

Is core clock = turbo clock? If yes maybe we need disable turbo clock. Maybe turbo cause this.

Maybe in PH II X4 L3 Cache helped a lot in multithreading efficiency. What make it wonderfull my 7.3GHz 6800K with NB almost 4ghz is not much faster than this, that with conditioner

Try AIDA64 Memory Test or MaxxMem please. Excavator look very good, even with less cache Nice info!

So this is not a benchmark, but score based on value you set that detected by XTU but must finish the test. What make me think this is not a benchmark because variable score is too tight, even if i test XTU with Youtube opened, score is not affected

So 743 is possible when 742 is possible. Good Job!

Really nice efficiency

Amazing Memory

Is this for real? Amazing

I got info Zen Opteron is 32 Core (16 + 16 in one die) with 8 Channel DDR4 Memory. So probably Zen Desktop will 16 Core with Quad Channel DDR4. And the most important i hope 40%+ IPC Improvement is from Excavator Core, not from older core. It will fun if memory controller is a beast like 4266Mhz Quad Channel for daily usage

After check @xxbassplayerxx BIOS, it exactly same function & OCP limit, so maybe both is ok for LN2

If you need more you can edit with VBE7, im newcomer at extreme cooling. So maybe other can help you. But as far as i know HD 7970 reference is about 225w, so 420w is a lot.

For those who want know BIOS specification, this is it OCP limiter is 420w @ +20%. I found this bios long time ago from other forum but i forgot which one. So i just upload it.

Is this what are you looking for? Hope it help

This is maybe hardest GFP lol!

Yep, your CPU help a lot, My GTX 480 with 2600K score much slower

Nice job & efficiency!!!

Little Update, still no see LOD effect, so i dont use LOD. Idk this is good result or not... Submission Link : http://hwbot.org/submission/3110955_speed.fastest_3dmark03_geforce_gtx_480_117790_marks

First Run Submission Link : http://hwbot.org/submission/3110609_speed.fastest_3dmark03_geforce_gtx_480_107159_marks

Craaazyyy! GJ!

I don't think any human run Prime95 for daily usage

Wow, really good idea! I think for GPU is better based on Codename because many rebranded card. If not based on GPU architecture name (GCN, Kepler, Maxwell, etc).

As far as i know, overclocking is not about raw performance, but performance improvement over stock speed, except for World Record is different. Overclock Athlon 5350 to 4ghz maybe is harder than do 6700K to 6.5ghz. Thats why HWBot give Hardware Point (pure point for overclocking) and Global Point (for fastest overclocked system). So to make 3D popular you need to accommodate what people need, cheap oc. With cheap oc people will submit more and make your 3D point will naturally get higher point, even more than XTU if submission become more popular.

Actually CPU can throttling if they exceeding max safe temperature. Different CPU different limit. And actually CPU really hard to do stability test on high temperature and get good score (that mean without throttling).