Search results

U
Core M v.s. A8X in Geekbench 3

As mentioned, I'd like to feedback formally through my perf team. GEMM has been around for decades, and there's an absolute swarm of literature out there about it, CPUs and GPUs alike. I need not google/IEEE Xplore for you. But to the point, how about an easy find from this year (2014)...
- UnmskUnderflow
- Post #92
- Oct 24, 2014
- Forum: CPUs and Overclocking
U
Core M v.s. A8X in Geekbench 3

I'm in hardware, so I'll get in touch with some trace guys who already have a channel to you through more official channels (preserving my anonymity here and not causing any conflicts of interest outside the job). I'll work with them on some suggestions. Anyway, I just noticed the GEMM format...
- UnmskUnderflow
- Post #88
- Oct 23, 2014
- Forum: CPUs and Overclocking
U
Core M v.s. A8X in Geekbench 3

@jfpoole Can I nitpick a bit? The GEMM kernel While functionally correct, this violates the 2 golden rules of GEMM 1.) Don't thrash the LS (2x loads per mul/accumulate) 2.) Don't have dependent mul/accumulate In a small enough memory footprint, this could give undue advantage to...
- UnmskUnderflow
- Post #61
- Oct 22, 2014
- Forum: CPUs and Overclocking
U
java test: unexpected results on AMD CPUs

In DrMrLordX's defense, if anything this shows what absurd crap can happen to simple loops in VMs or JITs. Overflow of int tanking perf? No easy IEEE 754 control? Weird as crap reactions to massive thread counts? His code is open source and is becoming more messy the more real it gets. This...
- UnmskUnderflow
- Post #70
- Sep 25, 2014
- Forum: CPUs and Overclocking
U
java test: unexpected results on AMD CPUs

For what it's worth, I tried looking up the quick way to force Java floats to round to infinity (or up, or ceiling, or away from zero, or anything besides nearest even). If you did this, you could keep the thread count small, and after it saturated, each addition would increment the LSB...
- UnmskUnderflow
- Post #64
- Sep 24, 2014
- Forum: CPUs and Overclocking
U
java test: unexpected results on AMD CPUs

No worries, HPC guys get that wrong all the time. Classic solutions since you just want linear number movement 1) change the round mode to round to infinity (easiest to keep this consistent) 2) try just adding the number to itself. That will scale, but has exponential side effects...you may...
- UnmskUnderflow
- Post #57
- Sep 24, 2014
- Forum: CPUs and Overclocking
U
java test: unexpected results on AMD CPUs

float1++ float2++ Ok, bothered me enough to post. If this is an abused int-land shorthand for "float1 + 1.0f" then after 24 bits of increment your floats will have the 1.0 land in sticky and be by default rounded down. Thus for most of the loop it will be moving the float nums by zero. A smart...
- UnmskUnderflow
- Post #55
- Sep 24, 2014
- Forum: CPUs and Overclocking

RESOURCES

Top Bottom

Search results

Core M v.s. A8X in Geekbench 3

Core M v.s. A8X in Geekbench 3

Core M v.s. A8X in Geekbench 3

java test: unexpected results on AMD CPUs

java test: unexpected results on AMD CPUs

java test: unexpected results on AMD CPUs

java test: unexpected results on AMD CPUs