Implement the getHashTimes() API #72
Conversation
Obtain the hash times mentioned in DLTcollab#47 with API.
| @@ -41,6 +41,7 @@ typedef struct { | |||
| cl_ulong max_memory; | |||
| size_t num_work_group; | |||
| KernelInfo kernel_info; | |||
| uint64_t hashTimes; | |||
| @@ -64,3 +64,8 @@ int8_t *getPoWResult(ImplContext *impl_ctx, void *pow_ctx) | |||
| { | |||
| return impl_ctx->getPoWResult(pow_ctx); | |||
| } | |||
|
|
|||
| uint64_t getHashTimes(ImplContext *impl_ctx, void *pow_ctx) | |||
There was a problem hiding this comment.
Why do you need implementation-specific counter?
There was a problem hiding this comment.
We need a counter in each pow_ctx.
There was a problem hiding this comment.
I'll set a counter named hash_count rather than hash_times to avoid misunderstanding in each pow_ctx.
And I'm not sure about the meaning of @jserv 's question.
There was a problem hiding this comment.
I would like to see the methods to minimize changes against existing interface. We might split the functions of given interface into two fields:
- ops (operations)
- info (information)
That is, PoW takes ops after initialization, which expects info.
| @@ -681,6 +681,22 @@ static int8_t *PoWAVX_getPoWResult(void *pow_ctx) | |||
| return ret; | |||
| } | |||
|
|
|||
| static uint64_t PoWAVX_getHashTimes(void *pow_ctx) | |||
There was a problem hiding this comment.
There are various lines in common among PoWAVX_getHashTimes, PoWC_getHashTimes, etc. I don't think we really need that.
There was a problem hiding this comment.
@marktwtn You can calculate the HashTimes inside the end of doThePoW function and store it. Then we can just use the interface get the HashTimes rather than duplicate getHashTimes functions in each implementation.
uint64_t getHashTimes(void *pow_ctx)
{
return pow_ctx->hashtimes;
}
There was a problem hiding this comment.
But the code would look like this
uint64_t getHashTimes(void *pow_ctx)
{
return ((PoW_SSE_Context *) pow_ctx)->hashtimes;
}
and can not be used in other implementation.
I guess the getHashTimes function should remain the same and change the PoW[Impl]_getHashTimes function like this:
static uint64_t PoW[Impl]_getHashTimes(void *pow_ctx)
{
return ((PoW_[Impl]_Context *) pow_ctx)->hashtimes;
}
| @@ -681,6 +681,22 @@ static int8_t *PoWAVX_getPoWResult(void *pow_ctx) | |||
| return ret; | |||
| } | |||
|
|
|||
| static uint64_t PoWAVX_getHashTimes(void *pow_ctx) | |||
There was a problem hiding this comment.
@marktwtn You can calculate the HashTimes inside the end of doThePoW function and store it. Then we can just use the interface get the HashTimes rather than duplicate getHashTimes functions in each implementation.
uint64_t getHashTimes(void *pow_ctx)
{
return pow_ctx->hashtimes;
}
| @@ -64,3 +64,8 @@ int8_t *getPoWResult(ImplContext *impl_ctx, void *pow_ctx) | |||
| { | |||
| return impl_ctx->getPoWResult(pow_ctx); | |||
| } | |||
|
|
|||
| uint64_t getHashTimes(ImplContext *impl_ctx, void *pow_ctx) | |||
There was a problem hiding this comment.
We need a counter in each pow_ctx.
|
|
||
| for (int i = 0; i < ctx->num_threads; i++) { | ||
| if (pitem[i].ret >= 0 ) | ||
| count += (uint64_t)pitem[i].ret; |
There was a problem hiding this comment.
Please follow the coding style, thanks.
count += (uint64_t) pitem[i].ret;
| if (pitem[i].ret >= 0 ) | ||
| count += (uint64_t)pitem[i].ret; | ||
| else | ||
| count += (uint64_t)(-(pitem[i].ret) + 1); |
|
|
Yes, it would be included in the other future pull request of getting the hash rate. |
|
I rewrite the code in #73 without implementing getHashTimes() API. If there is no problem, I will close this pull request without merging. |
Agree. |
Obtain the hash times mentioned in #47 with API.
The FPGA part will be implemented by @ajblane .