AB2012: ^ That stuff is mostly just marketing and the problem isn't that were "using old information", rather that SSD's have significantly changed the situation. ASUS's benchmarks when using 5,400rpm HDD's will show a dramatic improvement. But with SSD's not only are they much faster themselves, they also have far larger RAM caches inside them. Eg, today I have a 2TB MX500 that comes with a sizable 2GB DRAM cache that already does exactly the same thing at exactly the same capacity but without eating up any actual RAM. And Windows will still be caching stuff in unused RAM on top of that. Adding a 3rd layer of cache doesn't make much difference.

Example - we can already see 4x increase in benchmark speeds comparing SATA SSD's (550MB/s) to NVMe (+2GB/s) SSD's, so games must load 4x faster right? Here's how they scale in the real world:-
https://www.youtube.com/watch?v=V3AMz-xZ2VM
https://www.youtube.com/watch?v=YWtspZqOxAM

RAMDisks are even more into the realms of "depreciating gains" due to how games work - they'll request a chunk of data, unpack it into RAM, initialize some stuff, then request the next bit. They don't behave remotely like a synthetic CrystalDiskMark pure Sequential read speed test, and there's a whole lot of marketing selling ultra-premium Real Gamer (tm) branded stuff that promises you'll "feel" the difference based on blind number chasing 10x more than you actually will.

I think the problem is, most non AAA games don't use anywhere near 32GB and people who've bought themselves a ton of RAM often end up trying to find stuff to fill it up with, hoping that RAM acceleration software will make more of a difference that it does. The biggest beneficiaries of 32GB tend to be productivity (eg, video editing, databases, etc) rather than GOG games. The biggest beneficiaries of NVMe vs SATA SSD's tends to be 4K video editing / productivity more than games. And the biggest beneficiaries of RAMDisks tends to be database servers and "live boot" or virtualized OS's more than consumer game load times that are already being read from an SSD which reads faster than the games typically request chunks of data in a non-continuous manner.

Radiance1979: ...that is exactly the drive i used for testing total war three kingdoms, it does matter, from stutter less scrolling to faster turn times with and without the program, i sincererly mean to say that without using ramcache III scrolling the huge campaignmap on my system leaves me with micro stutter here and there

AstralWanderer: You shouldn't be seeing micro-stuttering with that type of system (and a 100MB Ram Cache would hardly be enough to make a difference). Check what else is running on your system (using the tools linked to above) and, in particular, try disabling any programs that modify your display (e.g. any overlays or benchmarking/FPS indicators).

If you're running Windows 10, try disabling memory compression - other users have highlighted this as a cause of micro-stuttering.

AB2012: If you have a decent SSD, then RAMDisks generally aren't worth it for gaming. First of all, you need to copy the game from disc into RAMDisk in the first place, which often takes as much time as loading it (actually more if you count the "human time" of setting it up and pre-loading it).

AB2012: like a CRC check on the game folder

Radiance1979: RAMCache lll software turns milliseconds into microseconds to boost game-load times.

rtcvb32: Starting to sound like another SoftRam....

AstralWanderer: Heh, the info about Windows 10 memory compression reminded me of another "memory booster" product from that era, QuarterDeck MagnaRAM, which also worked by compressing memory. Most reviews I read of it indicated that it caused greater system instability and that was back in the not-too-stable Win95 days.

GameRacer: Somewhat related....does anyone know if(on system with no ram expansion space left) using mem thumb drives as RAM(USB3 ones) is a good idea for games that need extra memory to run or to just run better?

AstralWanderer: USB (even 3.0) thumb drive performance has, in my experience, been pretty unimpressive - level pegging with an HDD on sequential access and far behind SSDs on random. So unless your main system is installed on a slow HDD, I'd suggest not. But let benchmarks be your guide.

An SSD connected via USB 3.0 should offer much better performance, but still far slower (by a factor of 10 at least) than RAM.

GameRacer: Well I was thinking more of being able to give some RAM intensive games(games using 8GB or even more of RAM) a bit more breathing room(since I cannot install more RAM).

GameRacer: Aside: In terms of speed, then, would it be better to just use paging file as extra RAM or use thumb drive style "ram"? Which would give better performance(HDD is 7200 RPM standard HDD afaik) of the two?

AstralWanderer: Heh, the info about Windows 10 memory compression reminded me of another "memory booster" product from that era, QuarterDeck MagnaRAM, which also worked by compressing memory. Most reviews I read of it indicated that it caused greater system instability and that was back in the not-too-stable Win95 days.

AstralWanderer: Best bet is to go through your system setup and prune back on those background processes using the most RAM. Disable unneeded Windows services - BlackViper is a useful source of information here - and use a utility like
Autoruns to check what you have running on startup.

AstralWanderer: Take a second look at what's available for your laptop RAM-wise - even if your existing SO-DIMM slots are filled, you may be able to install higher-density sticks (though this will mean replacing existing RAM).

AstralWanderer: Your paging file should already be being used as "extra RAM". If you are running on an HDD (most desktop ones are 5400RPM btw, so virtually all laptop ones will be too, if not less) then look at switching to an SSD to reduce the impact of heavy pagefile usage.

Thumb drives (presumably you are considering SuperFetch?) only provide a measureable benefit when you have a significant RAM shortage and are running off an HDD (with relatively large access times).