Change "lots" to 10 Terahashes

[fanquake]

No description provided.

[TheBlueMatt]

a Terahash is very unknown to the general public, can we convert it to estimated MIPS?

[fanquake]

Million Instructions per Second? I'm not to sure about the conversion though.

[TheBlueMatt]

Most recent conversion I've seen is https://bitcointalk.org/index.php?topic=4689.msg68933#msg68933 which states "one bitcoinhash is ~6.35k x86 INTOP" but it would be nice to ask a OCL kernel writer for an estimate of how many INTOPs they use per hash. Or you could phrase it to make it clear its "equivalent to N X86 MIPS"

[saivann]

Ok we have for x86 CPU : ~6350 INT OP / bitcoin hash
https://bitcointalk.org/index.php?topic=4689.msg68933#msg68933

Or we have for GPU : ~3385 INT OP / bitcoin hash
https://bitcointalk.org/index.php?topic=7964.msg550288#msg550288
http://bitcoin.stackexchange.com/questions/1293/how-many-integer-operations-on-a-gpu-are-necessary-for-one-hash

Do you think these values are enough accurate so that they can be used for an estimation? Otherwise, I'm willing to ask the right person to answer this question if you point me in the right direction.

I have very limited knowledge about these things, but from what I read, MIPS might not be the best unit to compare operations between different architectures. http://www.futuretech.blinkenlights.nl/perf.html

[gmaxwell]

A significant fraction of total mining power is now on FPGAs and ASICs which are even less comparable to x86 and GPUs than GPUs were to x86.

[saivann]

Absolutely. But can it still be relevant to display the relative equivalent processing power required by any traditional CPU or GPU to do the same operation? Ex. 60 Terrahashes/s ( equivalent to 762 petaFLOPS for x86 CPUs ) Or should we translate this differently?

[saivann]

Now discussed in issue #143