@MageTank @Prysin Care to get me started?

I have absolutely no input to offer on anything software related, simply because I do not have a great understanding of how programming works. That being said... 

I wouldn't mind something that could effectively benchmark either CPU or ram (or both) that scales with one or the other (or both) that offered a score, or "completion time" at the end of it. The easier to use/understand, the better. At the moment, all we really have to twiddle our thumbs with, are SuperPi and Cinebench.

Or, since you enjoy AVX, perhaps you can set out to create something hotter than Linpack itself? Granted, anything Linpack related gets hot if you throw enough ram at it, lol. I just don't know of anything to suggest that I would personally use, other than benching/stressing tools. For most other applications, as long as they work within the means of my hardware, I could care less how they are coded.

10 minutes ago, MageTank said:

I have absolutely no input to offer on anything software related, simply because I do not have a great understanding of how programming works. That being said... 

 

I wouldn't mind something that could effectively benchmark either CPU or ram (or both) that scales with one or the other (or both) that offered a score, or "completion time" at the end of it. The easier to use/understand, the better. At the moment, all we really have to twiddle our thumbs with, are SuperPi and Cinebench.

 

Or, since you enjoy AVX, perhaps you can set out to create something hotter than Linpack itself? Granted, anything Linpack related gets hot if you throw enough ram at it, lol. I just don't know of anything to suggest that I would personally use, other than benching/stressing tools. For most other applications, as long as they work within the means of my hardware, I could care less how they are coded.

PiFast is better than SuperPi, and there are many better Pi calculation benches out there today. As for something that scales with RAM, I'm sure I can arrange that fairly easily.

how about a setup that hashes 10 different big integers, then do a series of sqroot floats with FP64 and finally outputs the average number of all those values.

1 hour ago, Prysin said:

how about a setup that hashes 10 different big integers, then do a series of sqroot floats with FP64 and finally outputs the average number of all those values.

Define big integer. Arbitrary digit count?

The second one is pretty easy. That's just a 1D reduction. I'll get to work on the second one tomorrow evening after work. I haven't done encryption before, so that'll take some research.

41 minutes ago, patrickjp93 said:

Define big integer. Arbitrary digit count?

 

The second one is pretty easy. That's just a 1D reduction. I'll get to work on the second one tomorrow evening after work. I haven't done encryption before, so that'll take some research.

odd number integer of minimum 9 digits. can be any number. But the longer the number, the longer it takes to find the SqRoot. especially with FP64 float.

To make it more interesting, turn the average into a percentage average of the total initial sum....

Ofcourse, demand atleast 10 decimal digits accuracy of the float values with an incredibly narrow window of "good enough". Since floats can never be 100% true when it comes to odd's

6 hours ago, MageTank said:

I have absolutely no input to offer on anything software related, simply because I do not have a great understanding of how programming works. That being said... 

 

I wouldn't mind something that could effectively benchmark either CPU or ram (or both) that scales with one or the other (or both) that offered a score, or "completion time" at the end of it. The easier to use/understand, the better. At the moment, all we really have to twiddle our thumbs with, are SuperPi and Cinebench.

 

Or, since you enjoy AVX, perhaps you can set out to create something hotter than Linpack itself? Granted, anything Linpack related gets hot if you throw enough ram at it, lol. I just don't know of anything to suggest that I would personally use, other than benching/stressing tools. For most other applications, as long as they work within the means of my hardware, I could care less how they are coded.

this playlist will further explain things. Although the two video series DO overlap. MIT and UCBerkeley has posted lots of good videos.

This will explain mantle and the gist of low level APIs.

Info on Vulkan
 

C++ "speed lesson"  <- bit outdated but fundamentals is solid

OpenMP Multithreading "speed-lesson" <- this will allow you to understand just HOW fucking easy it is to do multithreading. (click link, Embedding of that playlist isnt allowed)

Looking into Assembly languages (this one is REALLY dry and hard to wrap your head around. but once you get the gist of it things get better)

Game engine development - iv'e just started into looking at this. Ill probably get through it soon enough.

@patrickjp93 will probably correct me if any of these videos are bad info sources. Some of them may be outdated, but none of them should be so outdated they cannot serve as a foundation for modern coding.

49 minutes ago, Prysin said:

odd number integer of minimum 9 digits. can be any number. But the longer the number, the longer it takes to find the SqRoot. especially with FP64 float.

 

To make it more interesting, turn the average into a percentage average of the total initial sum....

 

Ofcourse, demand atleast 10 decimal digits accuracy of the float values with an incredibly narrow window of "good enough". Since floats can never be 100% true when it comes to odd's

Percentage average of the initial total sum...
 

Example?

say you have 10x9. So that is 90

90 = 100%

then you do the square root of 9 which is 3. 3x10 = 30. Which is 33.33333333% of the initial sum.

except the 10 initial integers need to be just random buttonsmashing odds.

@Prysin

What do you want as the hash function?

6 hours ago, patrickjp93 said:

@Prysin

What do you want as the hash function?

no specific idea, just make it complicated but readable.

53 minutes ago, Prysin said:

no specific idea, just make it complicated but readable.

Because when you say hash, I'm thinking hash table hashing. If you have a different idea, now's the time.

15 minutes ago, patrickjp93 said:

Because when you say hash, I'm thinking hash table hashing. If you have a different idea, now's the time.

no that was my idea too

30 minutes ago, Prysin said:

no that was my idea too

Okay.

2 minutes ago, patrickjp93 said:

Okay.

also, as a unrelated sidenote. I will start reading into Vulkan programming and how to actually program in a Windows enviroment sometimes next year... So for now you got a head-start.