ISRT refuses to see my Samsung SM951 128GB M.2 drive

[Frick]

Communication is so fascinating (especially when Pill Monster is involved).

PrimoCache sound really interesting, might have to look it up.

 

[Tatty_Two]

One or two of you seem to enjoy once again completely hijacking a thread and shifting the emphasis to personal arguments containing insults, so I will say this just this once, Stop it now and get back to discussing the specifics of the Op, anymore of this crap and my patience will have abandoned me for good, gentlemen......... take it to PM!!! ...... thank you.

 

[Pill Monster]

You guys are arguing in circles. Simple fact is that @newtekie1 and @RejZoR are right about this one. PrimoCache appears to use a kernel mode driver to intercept I/O requests from the disk. This means at this point, Windows file cache has already been missed and is going forward to get the data from the disk. Primo is simply caching to the SSD and utilizing it when the I/O requests fall through to the disk. It sits between Windows file cache and your spinning disk as a kernel mode driver. It's not a replacement for Windows file cache, it's just another level below it.

@Pill Monster : I would stop fighting this one as you're not only wrong, you're becoming belligerent. I would rather you concede that you misunderstood what they were talking about and leave it at that because your posts clearly indicate that you're missing the point that they're making.

Simple fact is that the cache levels in PrimoCache are not the same thing as CPU cache. It's just confusing because Primo used the same names which only makes matters worse. I would actually read up on the very thing you're saying it isn't...

https://www.romexsoftware.com/en-us/primo-cache/term-l2cache.html

Spoiler

Hey man I totally missed the last part of this post yesterday, I wish I'd seen it but I was getting pretty frustrated and prob skipped it.

What you say could well be true, it may be able to cache to SSD as well, however that makes my point even more valid.

Windows already caches files to RAM, this is why I posted the screenshot yesterday with memory mapped files from Panzer Corps displayed. Those are cached files sitting in physical RAM, after opening Panzer Corps then closing it again.
It's a rough example, but one I thought was clear... Windows will keep on caching to RAM until it either runs out of space or room is needed for higher priority pages. Cache doesn't count against the system Commit Charge.

So instead of just doing the same job as Windows, Primo actually does a worse job because it's caching to SSD when Windows is already caching the files to RAM. Hence the snakeoil.

Primo has been around for years...I've been involved in a few discussions regarding it's effectiveness. I can even link you to Guru3D where Hilbert eventually chimed in and agreed it was no better than Windows' own caching.



Tbh I'm starting to think there's something wrong with my communication skills.....I can't figure out exactly what though. I wish I knew....seriously.

Using spoiler to avoid thread hijacking.

 

[RejZoR]

Oh my god. OH MY GOD. You have to populate RAM cache on EVERY system reboot. SSD cache is ready at any time. Why do people buy SSD drives if you can have HDD and it's anyway all cached to RAM? See, that logic doesn't float for long.

And while I have 32GB of RAM, most users have only like 8GB. Not enough to cache 2-3 modern games entirely and still have all the RAM available. With SSD cache, you can. I really don't understand what's so difficult to understand about SSD caching. It's a HDD that becomes nearly as fast as SSD. For lower price and higher capacity with limitation that it doesn't work straight away like SSD, but it needs 1 or 2 runs of the app first and then it's boosted permanently.

 

[squeezehat]

He always talks about B when everybody else talks about A.

Also memory mapped files don't mean files that are cached. A memory mapped file means a file that is loaded into memory by addressing it like addressing memory - or modified/saved, if you write to the mapped area.