Should i buy primocache?

[ThrashZone]

Hi,
Superficial feature at best
How poor your system is and how much ram you have is the only possible perk
To me as a buy, it's a no and I already have similar included with my boards and still don't use it.

 

[BSim500]

You are missing important context and as I said... ...so move along.

LOL. No problem. By all means continue to be 'that guy' who 8 years said "Whoa man look at this DOUBLING IN SPEED improvement from turning 'RAPID' mode on in Samsung Magician", followed up by "But this time around, I'm not gonna bother testing to see if that translates to a 12.6s halving in game load times to 6.3s based on false extrapolation of the same synthetic maths in case it doesn't tell me what I want to hear..."

Seriously, I've been playing with RAMDisks since hacking DEVICE=C:\RAMDRIVE.SYS into CONFIG.SYS in MS-DOS. They have their uses, but they are not a "magic beans" way of tripling your SSD speeds for free in actual real-world software outside of the CrystalDiskMark bubble the way some marketing brochures portray...

 

[Shrek]

No. Waste of money. It's just a fancy RAM-Drive type software. Unless you have a frak-ton of RAM(128GB+) is will not be worth the purchase.

I was wrong. See testing below...

You gained my respect; few people these days seem able to admit they are wrong. I will be taking your posts all the more seriously.

 

[ThrashZone]

Hi,
Yeah almost forgot sammy's rapid mode voodoo feature I don't use either
So much superficial performance I missing that is free with those products man my benchmarks are really suffering

Money is money but waste is also waste you just have to pick what is either by yourself.

 

[newtekie1]

but not much in write speeds, which were only slightly better.

Your differences in write speeds are just margin of error. Primocache doesn't improve write speeds unless you enable Defer-Write. Defer-Write is disabled by default with a RAM cache because there is a risk of data loss if the system is powered off before the written data that is in the RAM cache is actually written to the drive.

I've used Primocache for years now, but I don't use the RAM cache. It just doesn't make a lot of sense to me. I just use it to add a large SSD cache to hard drive storage. IMO, that is a far more useful function, and you can enable defer-write without much risk of data loss from a power outage.

 

[Shrek]

I've used Primocache for years now, but I don't use the RAM cache. It just doesn't make a lot of sense to me. I just use it to add a large SSD cache to hard drive storage. IMO, that is a far more useful function, and you can enable defer-write without much risk of data loss from a power outage.

Given the way Solid State Drives can fail without warning, this seems like a great way to use them.

 

[TheoneandonlyMrK]

I tried all sizes (starting with 2GB) and ended up at 16-24GB (for caching sub 10GB smaller games outright). It can make a huge difference if you're loading from a 5,400rpm HDD and the algorithm gets it right, but vs a fast SSD, it didn't made much difference in the real-world for reasons mentioned. My RAMDisk shows CrystalDiskMark Sequential screenshots of over 12,000MB/s vs only 520MB/s from an MX500 SATA SSD, yet that really doesn't mean games will load 23x faster as games simply don't load like that. In the real world, they'll request a chunk of data, unpack it, initialise it, etc, then request another chunk. It isn't a continuous "sequential" stream at all. "If my SSD is 2,000MB/s and my RAMDisk is 10,000MB/s then that means game load times will fall from 15s to 3s" as some believe just doesn't happen in the real world for the same reason the newest P31 NVMe review isn't showing 6x lower load times for games than 2016-era SATA's on the same chart. The theoretical CrystalDiskMark Sequential -> real-world Game Load times never was something that ever scaled linearly, let alone do so to infinity beyond depreciating gains.

I agree.

Your differences in write speeds are just margin of error. Primocache doesn't improve write speeds unless you enable Defer-Write. Defer-Write is disabled by default with a RAM cache because there is a risk of data loss if the system is powered off before the written data that is in the RAM cache is actually written to the drive.

I've used Primocache for years now, but I don't use the RAM cache. It just doesn't make a lot of sense to me. I just use it to add a large SSD cache to hard drive storage. IMO, that is a far more useful function, and you can enable defer-write without much risk of data loss from a power outage.

This is how I used mine , the storage tiering is brilliant, better than AMDs free one, I swapped to all nvme then ditched it but this is a viable use of primo, I also used cache at the same time.
GtaV did load quicker but not quickly still.

 

[newtekie1]

This is how I used mine , the storage tiering is brilliant, better than AMDs free one, I swapped to all nvme then ditched it but this is a viable use of primo, I also used cache at the same time.
GtaV did load quicker but not quickly still.

At this point I only have my 8TB games drive in my main PC cached with a 1TB SATA SSD, and I have my RAID5 array in my home's media server cached with a 1TB NVMe SSD.

Edit: Here are some results from a RAM cache I through together just now for testing.

Baseline:

Cache w/o defer-write:

Cache w/ defer-write:

 

[Bill_Bright]

Interesting results Lex. Thanks for posting that.

However, I wonder if the performance gains are actually noticeable in real-world use? If using a HD, for sure. But if someone already has a big chunk of RAM, and a decent SSD, are they really going to see and experience any significant performance gains? I still have my doubts.

 

[newtekie1]

Interesting results Lex. Thanks for posting that.

However, I wonder if the performance gains are actually noticeable in real-world use? If using a HD, for sure. But if someone already has a big chunk of RAM, and a decent SSD, are they really going to see and experience any significant performance gains? I still have my doubts.

In my experience, using a RAM cache with a decent NVMe SSD makes no noticeable difference.

 

[Bill_Bright]

In my experience, using a RAM cache with a decent NVMe SSD makes no noticeable difference.

Same here. That's my point. It is like many so called enhancements - they only look better on paper and in benchmark programs. Good, maybe, for bragging rights, but they provide zero "noticeable" performance gains when compiling databases, downloading a movie, rendering a webpage or graphics image, spell checking a 300 page document, or playing a game.

 

[DeathtoGnomes]

@lexluthermiester A real test is moving 20,000+ music files, 100+gb, from one SSD to another.

 

[lexluthermiester]

You gained my respect; few people these days seem able to admit they are wrong. I will be taking your posts all the more seriously.

Despite what the perspective is, I have no problem admitting when I'm wrong and couldn't care less where the information comes from, whether my own discovery or someone else proving a point.

GtaV did load quicker but not quickly still.

Tried this and a few other games that load large amounts of data. I'm not seeing a difference with and without the cache on GTA5, but with Wolfenstein 2 & the Metro games I see a 15% to 20% drop in load times. Then tried...

@lexluthermiester A real test is moving 20,000+ music files, 100+gb, from one SSD to another.

...doing this. I copied files from one cached drive to a non-cached drive and vice-verse. Nothing. Not a bit of improvement.

This is making me wonder how this program works and why it performs well in some situations and not others..

 

[silentbogo]

I'm not going to argue. Believe what you want.

Somehow I missed the historical point where caching mechanisms became a religion....

 

[TheoneandonlyMrK]

Despite what the perspection is, I have no problem admitting when I'm wrong and couldn't care less where the information comes from, whether my own discovery or someone else proving a point.


Tried this and a few other games that load large amounts of data. I'm not seeing a difference with and without the cache on GTA5, but with Wolfenstein 2 & the Metro games I see a 15% to 20% drop in load times. Then tried...

...doing this. I copied files from one cached drive to a non-cached drive and vice-verse. Nothing. Not a bit of improvement.

This is making me wonder how this program works and why it performs well in some situations and not others..

One time transfers are not going to improve with a cache?!.

Regularly used files see improvement over time (not a long time but not instantly).

 

[ThrashZone]

One time transfers are not going to improve with a cache?!.

Regularly used files see improvement over time (not a long time but not instantly).

Hi,
Yeah this is the benchmark voodoo
Run it once and repeat benchmark improves second/... time.

 

[newtekie1]

Same here. That's my point. It is like many so called enhancements - they only look better on paper and in benchmark programs. Good, maybe, for bragging rights, but they provide zero "noticeable" performance gains when compiling databases, downloading a movie, rendering a webpage or graphics image, spell checking a 300 page document, or playing a game.

Yeah, it's why I don't use a RAM cache. Just use a fast NVMe drive(or heck even a SATA SSD if your system doesn't have NVMe) for your system drive. Where Primocache helps out is when you still have hard drives that you use regularly, or even run games/programs from.

I copied files from one cached drive to a non-cached drive and vice-verse. Nothing. Not a bit of improvement.

This is making me wonder how this program works and why it performs well in some situations and not others..

This is very much going to depend on how you have Primocache setup and the drives involved. If you have defer-write disabled(again it's disabled by default) then Primocache is not really going to help at all with file transfers. Lets just say you have the following setup:

Drive A + Cache
Drive B

With defer-write off, any data written to drive A will be written directly to the drive at the same time it is written to the cache. Writing is cached but not sped up. I know that sounds kind of stupid, but it does make some sense. The written data is stored in the cache so that if it is accessed for reading the reading will be significantly faster because the data is already in the cache.

So, with defer-write off, write speed to Drive A won't be any faster. Likewise, copying data from Drive A to Drive B also won't be any faster, since Drive B's write speed will likely be the limiting factor here. So copying data from A to B or vice-versa will probably not be any faster.

One time transfers are not going to improve with a cache?!.

It depends on how you have it setup. Yes, one time transfers can be improved with a cache if configured to do so.

For example, my RAID5 arrays in my server have terribly slow write speeds. Once the small amount of cache on the RAID controller is used up, write speeds just nosedive to below 100MB/s. But using Primocache with an SSD as a cache drive write speeds are way faster.

 

[TheoneandonlyMrK]

Yeah, it's why I don't use a RAM cache. Just use a fast NVMe drive(or heck even a SATA SSD if your system doesn't have NVMe) for your system drive. Where Primocache helps out is when you still have hard drives that you use regularly, or even run games/programs from.


This is very much going to depend on how you have Primocache setup and the drives involved. If you have defer-write disabled(again it's disabled by default) then Primocache is not really going to help at all with file transfers. Lets just say you have the following setup:

Drive A + Cache
Drive B

With defer-write off, any data written to drive A will be written directly to the drive at the same time it is written to the cache. Writing is cached but not sped up. I know that sounds kind of stupid, but it does make some sense. The written data is stored in the cache so that if it is accessed for reading the reading will be significantly faster because the data is already in the cache.

So, with defer-write off, write speed to Drive A won't be any faster. Likewise, copying data from Drive A to Drive B also won't be any faster, since Drive B's write speed will likely be the limiting factor here. So copying data from A to B or vice-versa will probably not be any faster.



It depends on how you have it setup. Yes, one time transfers can be improved with a cache if configured to do so.

For example, my RAID5 arrays in my server have terribly slow write speeds. Once the small amount of cache on the RAID controller is used up, write speeds just nosedive to below 100MB/s. But using Primocache with an SSD as a cache drive write speeds are way faster.

Fair point I should have been specific ,I meant in his test but your right to point out wrote caching could help.

 

[aliceif]

Now that I think of it, maybe it is significantly more effective on something like a DRAM-less drive with 90%+ used space since that would reduce its internal SLC cache, could be an interesting scenario to try.​

 

[newtekie1]

Fair point I should have been specific ,I meant in his test but your right to point out wrote caching could help.

Just to illustrate my use I ran a file copy test. I just crabbed a few Windows ISOs amounting to about 28GB and put them on OS SSD. The OS SSD is just a standard SATA SSD, so max read speed is about 500MB/s. I then copied them to my RAID5 array(this array is 3x12TB hard drives).

This is the beginning of the file transfer when the RAID controller's cache is being used and Primocache is turned off:

This is what happens after the RAID controller's cache is full and Primocache is still disabled:

And this is when I enabled Primocache:

 

[lexluthermiester]

One time transfers are not going to improve with a cache?!.

Regularly used files see improvement over time (not a long time but not instantly).

I thought about that. Exited and reloaded several games a few times to be sure. GTA5 saw no benefit at all. Not sure why. The other games I tested with and without the cache seemed to have a steady benefit. I was going to post each of FF14 Benchmarks but they ended up not showing any benefit either.

I'll admit to being a bit stumped. Why is this working for some situations and not others? Is it a glitch in the program? Am I missing a setting that is important? Clearly there are indications of potential...
This thing is intriguing but I've got a busy schedule. Anyone else want to do testing and see if we can figure it out?

 

[TheoneandonlyMrK]

I thought about that. Exited and reloaded several games a few times to be sure. GTA5 saw no benefit at all. Not sure why. The other games I tested with and without the cache seemed to have a steady benefit. I was going to post each of FF14 Benchmarks but they ended up not showing any benefit either.

I'll admit to being a bit stumped. Why it this working for some situations and not others? Is it a glitch in the program? Am I missing a setting that is important? Clearly there are indications of potential...
This thing is intriguing but I've got a busy schedule. Anyone else want to do testing and see if we can figure it out?

I would offer to help but the weekends rebuild is lingering on , it's built but flakey as Shiite ATM.
I used it (primo)solidly for a year on two 3TB drive's in raid 0 with a 256GB sata3 ssd as a cache and memory cache turned on, as you say it did improve load times but after swapping to 3Xnvme the in use gains were not apparent, they still showed in benches.
GtaV did load quicker for me but I'm talking a minute instead of three(obviously those are feeling not fact time's I didn't clock it)

 

[newtekie1]

I thought about that. Exited and reloaded several games a few times to be sure. GTA5 saw no benefit at all. Not sure why. The other games I tested with and without the cache seemed to have a steady benefit. I was going to post each of FF14 Benchmarks but they ended up not showing any benefit either.

I'll admit to being a bit stumped. Why it this working for some situations and not others? Is it a glitch in the program? Am I missing a setting that is important? Clearly there are indications of potential...
This thing is intriguing but I've got a busy schedule. Anyone else want to do testing and see if we can figure it out?

If the bottleneck isn't the storage device, then increasing the speed of the storage device won't help. I've seen GTAV not benefit from going from a HDD to a SSD. The initial loading screen just takes forever, I'm not sure what it is doing in the background, I suspect it is de-compressing game assets during that part.

 

[RAA!]

Ohh this makes me want to purchase one of those T3610's off ebay for cheap and stuff it with ram.

 

[lexluthermiester]

Before I respond, I couldn't see your posts. Resolved.

With defer-write off, any data written to drive A will be written directly to the drive at the same time it is written to the cache. Writing is cached but not sped up.

Interesting, I had not used that function.

I know that sounds kind of stupid, but it does make some sense. The written data is stored in the cache so that if it is accessed for reading the reading will be significantly faster because the data is already in the cache.

Actually, that makes perfect sense and is logical. I think it's time to try again..