Guru3D.com Forums

Go Back   Guru3D.com Forums > Hardware > SSD and HDD storage
SSD and HDD storage This section covers storage hardware like SSDs HDDs, RAID chatter and everything related


Reply
 
Thread Tools Display Modes
FancyCache / PrimoCache
Old
  (#1)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Lightbulb FancyCache / PrimoCache - 08-06-2013, 23:42 | posts: 207 | Location: England

Quote:
PrimoCache is a supplementary software caching scheme that cooperates with system memory to provide data caching for disks. It improves system performance by transparently storing data into memory such that future requests for that data can be served faster. PrimoCache caches data on a logical block basis (offsets within a disk) while windows cache manager caches on a virtual block basis (offsets within a file).

PrimoCache can also utilize the OS Invisible Memory which is wasted on 32-bit Windows when 4GB memory or more is installed.

Principle

One of the PrimoCache's core components is a storage class filter driver which resides in the storage stack, intercepting I/O requests for data on disk. If requested data is contained in the cache, this request can be served by simply reading the cache, which is comparably faster. Otherwise the data has to be fetched from disk. Hence, the more requests can be served from the cache the better the overall system performance is.
More Info here.

I haven't seen any post about this program, for me it seems pretty useful when you got extra ram to burn,

Download

Here some benchmarks using it with my SSDs.


This is what the program looks like when setting up caches for HDD/SSD

Last edited by holystarlight; 08-07-2013 at 01:29.
   
Reply With Quote
 
Old
  (#2)
tsunami231
Ancient Guru
 
tsunami231's Avatar
 
Videocard: EVGA 660gtx sig2
Processor: i7 920 CNPS10X Quiet
Mainboard: Evga x58 SLI LE
Memory: 3x2gb Dominator@1600 6Gb
Soundcard: Realtek HD Audio
PSU: Antec Truepower 750
Default 08-07-2013, 01:15 | posts: 3,201 | Location: USA

Ramdisk, nothing new about them
   
Reply With Quote
Old
  (#3)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-07-2013, 01:27 | posts: 207 | Location: England

Quote:
Originally Posted by tsunami231 View Post
Ramdisk, nothing new about them
It Actually works very differently to a ramdisk. it isn't used to store files at all, works more like a buffer to the SDD/HDD by dedicating a ram cache, Just like how mechanical HDD use 64mb cache buffer for example.
   
Reply With Quote
Old
  (#4)
tsunami231
Ancient Guru
 
tsunami231's Avatar
 
Videocard: EVGA 660gtx sig2
Processor: i7 920 CNPS10X Quiet
Mainboard: Evga x58 SLI LE
Memory: 3x2gb Dominator@1600 6Gb
Soundcard: Realtek HD Audio
PSU: Antec Truepower 750
Default 08-08-2013, 20:58 | posts: 3,201 | Location: USA

its ramdisk that only way ones gets those kind read/writes speeds
   
Reply With Quote
 
Old
  (#5)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-08-2013, 21:41 | posts: 207 | Location: England

Quote:
Originally Posted by tsunami231 View Post
its ramdisk that only way ones gets those kind read/writes speeds


Doesn't look like any ramdisk iv used, as you can see, from the image running a benchmark on my C: drive, as I explained before, its works more like a buffer than a ramdisk. anyway, gotta say this program does do some magic when transferring files from one HDD to another, and access times have improved greatly.
   
Reply With Quote
Old
  (#6)
Hilbert Hagedoorn
Don Vito Corleone
 
Hilbert Hagedoorn's Avatar
 
Videocard: AMD | NVIDIA
Processor: Core i7 4770K
Mainboard: Z77
Memory: 8GB
Soundcard: X-Fi - GigaWorks 7.1
PSU: 1200 Watt
Default 08-11-2013, 19:38 | posts: 20,474 | Location: Guru3D testlab

It's a RAMCache not a logical ramdisk, so yes this functions as a buffer.

I'd be a bit careful though with power loss / unexpected resets / crashes as critical data will get lost when that happens. Other then that it indeed is that fastest cache you can create.


Follow Guru3D on twitter.
Follow Guru3D on facebook.
   
Reply With Quote
Old
  (#7)
Pill Monster
Ancient Guru
 
Pill Monster's Avatar
 
Videocard: 7950 Vapor-X 1175/1550
Processor: AMD FX-8320 @4.8
Mainboard: ASUS Sabertooth 990FX R2
Memory: 8GB Kingston HyperX 2400
Soundcard: Audigy 2 Platinum Ex 5.1
PSU: AcBel M8 750
Default 08-11-2013, 20:57 | posts: 23,677 | Location: NZ

Quote:
Originally Posted by holystarlight View Post

Doesn't look like any ramdisk iv used, as you can see, from the image running a benchmark on my C: drive, as I explained before, its works more like a buffer than a ramdisk. anyway, gotta say this program does do some magic when transferring files from one HDD to another, and access times have improved greatly.
I really doubt that considering your write speeds don't change - only read speed and access times.

Fancy Cache is basically Superfetch with a few extra options.....

Last edited by Pill Monster; 08-11-2013 at 21:01.
   
Reply With Quote
Old
  (#8)
FatBoyNL
Maha Guru
 
FatBoyNL's Avatar
 
Videocard: MSI GTX770 TFG 1320/7712
Processor: i7-2600K 45x100 Mugen2B
Mainboard: Asus P8Z68 Deluxe 3603
Memory: 4x4GB GSkill Sr-F3 1866C9
Soundcard: ZxR|RX-A1020 DT880|MT-60D
PSU: CM Silent Pro Gold 800W
Default 08-13-2013, 14:13 | posts: 1,297 | Location: The Netherlands

With 1 run of 50MB the HDD or SSD is not even used at all.
Please try it again with multiple runs of blocks larger than the cache size. For example, test 4x 500MB. I wonder what results that would show.
   
Reply With Quote
Old
  (#9)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-13-2013, 22:23 | posts: 207 | Location: England

Quote:
Originally Posted by FatBoyNL View Post
With 1 run of 50MB the HDD or SSD is not even used at all.
Please try it again with multiple runs of blocks larger than the cache size. For example, test 4x 500MB. I wonder what results that would show.
ill test
Without Cache 4x 500mb


with cache 4x 500mb


And with 4x 2000mb


4x 4000mb


And with AS SSD bench


Very impressive results, Cache size I'm using is 4096 with block size 16 on a SSD OCZ Agility 3.

Last edited by holystarlight; 08-14-2013 at 01:41. Reason: Added more benchmarks.
   
Reply With Quote
Old
  (#10)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-14-2013, 00:03 | posts: 207 | Location: England

Quote:
Originally Posted by Pill Monster View Post
I really doubt that considering your write speeds don't change - only read speed and access times.

Fancy Cache is basically Superfetch with a few extra options.....


this is copying a 3.6gb ISO image from one HDD to another both HDD using a 4096 cache with block size 16. the file copies so quickly that kinda hard to screenshot it lol. so I did another test.



Here is test done with and without the cache copying the same file to and from same locations, first copy is with the cache on, and the second is without.





Also to note that primoCahce uses something called Defer-Write with a latency delay, which is customizable.

Quote:
Defer-Write: If ticked, incoming write-data will not be immediately written to disk. Write-data is first stored in cache and then written to disk after certain delays. Defer-Write will greatly improve write performance. However, a power outage or system failure might result in data loss or corruption. It is recommended that users only enable Defer-Write on volumes where temporary, unimportant or reproducible data is to be stored.

Last edited by holystarlight; 08-14-2013 at 01:42.
   
Reply With Quote
 
Old
  (#11)
FatBoyNL
Maha Guru
 
FatBoyNL's Avatar
 
Videocard: MSI GTX770 TFG 1320/7712
Processor: i7-2600K 45x100 Mugen2B
Mainboard: Asus P8Z68 Deluxe 3603
Memory: 4x4GB GSkill Sr-F3 1866C9
Soundcard: ZxR|RX-A1020 DT880|MT-60D
PSU: CM Silent Pro Gold 800W
Default 08-14-2013, 20:02 | posts: 1,297 | Location: The Netherlands

Those results look promising!
But the improved reads will only shine when the data to be fetched has been cached already. A 40GB SSD cache drive using Intel Smart Response Technology will cache the most accessed data 'more permanently' (but a lot slower).
PrimoCache probably does the same thing by software in RAM. Shutdows will be slower as the data has to be written to the HDD and startups will be slower due to starting caching the data from the disk. In the background probably.

About the writes. When you copy a very large file from G- to C-drive, at what speed does the action end? The file still needs to be copied to the slower HDD, so I reckon the speed rate will fall dramatically at the end of the copying action (or at shutdown or other checkpoints).
Please show us a screenshot of a 10GB copying action over those drives at 99% completion

I'm just kidding, I'm very interested about this topic. But from my (professional) experience, it's very hard to notice any improvement except for specific situations (which I haven't found yet really ).

Last edited by FatBoyNL; 08-14-2013 at 20:05.
   
Reply With Quote
Old
  (#12)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-14-2013, 21:39 | posts: 207 | Location: England

Quote:
Originally Posted by FatBoyNL View Post
Those results look promising!
But the improved reads will only shine when the data to be fetched has been cached already. A 40GB SSD cache drive using Intel Smart Response Technology will cache the most accessed data 'more permanently' (but a lot slower).
PrimoCache probably does the same thing by software in RAM. Shutdows will be slower as the data has to be written to the HDD and startups will be slower due to starting caching the data from the disk. In the background probably.

About the writes. When you copy a very large file from G- to C-drive, at what speed does the action end? The file still needs to be copied to the slower HDD, so I reckon the speed rate will fall dramatically at the end of the copying action (or at shutdown or other checkpoints).
Please show us a screenshot of a 10GB copying action over those drives at 99% completion

I'm just kidding, I'm very interested about this topic. But from my (professional) experience, it's very hard to notice any improvement except for specific situations (which I haven't found yet really ).
The program does have a L2 cache that can be used on a dedicated SSD like how Intel Smart Response Technology works.
Quote:
Level-2 Cache: Uses SSD, flash drive, or other faster device as a secondary cache which will speed up traditional slow hard disks. Cache contents stored in level-2 storage are persistent across computer restarts. Users must not "offline" modify the contents of a volume which is being cached by level-2 cache.
Level-2 Storage: A volume formatted to provide cache and store cached data. To create a level-2 storage, click "Create/Manage Level-2 Storages" button. Note: only partitions on Basic MBR disks are supported to be level-2 storages.
Also you can choose which Algorithm the program uses
Quote:
Algorithm: Determines which blocks to discard to make room for new data when cache is full.
LRU (Least Recently Used): Discards the least recently used data first.
LFU-R (Least Frequently Used): Counts how often a data block is needed. Data blocks that are used least often will be discarded first.
but I haven't a spare SSD to test it out the L2 Cache, but I haven't noticed any slow shut downs or start up, Task manager doesn't show the program running in the background either,

Ill test copying a larger file like 10gb but I haven't a 10gb to test with so ill make one and post a bench or two.

with cache

And without

Last edited by holystarlight; 08-14-2013 at 22:28.
   
Reply With Quote
Old
  (#13)
Chillin
Ancient Guru
 
Chillin's Avatar
 
Videocard: Gigabyte GTX 560 930/2300
Processor: i5-2500K@4.4GHz 1.2v H60
Mainboard: Asrock Z77 Pro4
Memory: G.Skill 2X4GB DDR3-1600
Soundcard: X-Fi XtremeGamer+Z506 5.1
PSU: Corsair TX 750w v2
Default 08-24-2013, 11:39 | posts: 6,446 | Location: Chilling

While I'm usually skeptical about such programs, this one seems to be the real deal.

I have set up 512mb of cache to read only (I don't trust writes with a power failure risk) on a WD 1TB Caviar Black (first gen) and here are the results:

Without:


With:


I'm going to see if I see any difference in BF3 map loading time.
   
Reply With Quote
Old
  (#14)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-24-2013, 18:17 | posts: 207 | Location: England

Quote:
Originally Posted by Chillin View Post
While I'm usually skeptical about such programs, this one seems to be the real deal.

I have set up 512mb of cache to read only (I don't trust writes with a power failure risk) on a WD 1TB Caviar Black (first gen) and here are the results:

Without:


With:


I'm going to see if I see any difference in BF3 map loading time.
yeh it works a treat, for HDD, using read only mode like you, for my storage HDDs, iv noticed tomb raider loads a faster, but most games I play already load pretty quickly so hard to notice any improvements.
   
Reply With Quote
Old
  (#15)
sykozis
Ancient Guru
 
sykozis's Avatar
 
Videocard: eVGA GTX660SC SLI
Processor: Core i7 2600K
Mainboard: ASRock Z77 Extreme4
Memory: 8gb G.Skill DDR3-1866
Soundcard: Creative Recon3D PCIe
PSU: SeaSonic M12II 620 Bronze
Default 08-24-2013, 19:12 | posts: 16,075 | Location: US East Coast

I decided to give this a try. I setup a 512mb read "cache" for my data drive. Copied files ranging from 700mb to 4.5GB. On the 700mb files, it was roughly 1-2 sec faster. On the 4.5GB file, it was noticeable slower. I retested the 4.5GB file with a 1024mb read "cache" with no improvement. Since I tend to transfer several GB of data from drive to drive on a regular basis...this software appears to be pretty useless for me. Since most harddrive benchmarks are synthetic, their results are meaningless. According to the data transfer progress window, the data transfer rate was higher with this software.....but the actual data transfer TIME was noticeably slower with my 4.5GB file (which is actually the full OpenSuSE x64 disc ISO).


   
Reply With Quote
Old
  (#16)
holystarlight
Master Guru
 
holystarlight's Avatar
 
Videocard: MSI 780 lightning SLI
Processor: i7 4960x 4.9Ghz
Mainboard: Asus Rampage 4 BE
Memory: Corsair Platinum 2400mhz
Soundcard: Asus Xonar Phoebus
PSU: Corsair AX1200i
Default 08-24-2013, 19:41 | posts: 207 | Location: England

Quote:
Originally Posted by sykozis View Post
I decided to give this a try. I setup a 512mb read "cache" for my data drive. Copied files ranging from 700mb to 4.5GB. On the 700mb files, it was roughly 1-2 sec faster. On the 4.5GB file, it was noticeable slower. I retested the 4.5GB file with a 1024mb read "cache" with no improvement. Since I tend to transfer several GB of data from drive to drive on a regular basis...this software appears to be pretty useless for me. Since most harddrive benchmarks are synthetic, their results are meaningless. According to the data transfer progress window, the data transfer rate was higher with this software.....but the actual data transfer TIME was noticeably slower with my 4.5GB file (which is actually the full OpenSuSE x64 disc ISO).
I think you will only notice an improvement with the write cache on aswell, as read mode shouldn't really help with transferring files from one place to another.
but iv been using an 8gb cache. As I got endless amount of ram, and I have noticed when copying a file larger than the cache, it will max out the cache then the latency kicks in to flush the cache to the HDD/SDD.
so from my testing you get a burst of speed and if the cache maxes out, the transfer speed drops, as the cache is flushed to the HDD, then it spikes back up again. but using a large cache seems to limit that issue.

Playing around with the latency option might help along with a larger cache.

that's what I have engaged anyway, I might be mistaken.

Last edited by holystarlight; 08-24-2013 at 19:43.
   
Reply With Quote
Old
  (#17)
Pill Monster
Ancient Guru
 
Pill Monster's Avatar
 
Videocard: 7950 Vapor-X 1175/1550
Processor: AMD FX-8320 @4.8
Mainboard: ASUS Sabertooth 990FX R2
Memory: 8GB Kingston HyperX 2400
Soundcard: Audigy 2 Platinum Ex 5.1
PSU: AcBel M8 750
Default 08-24-2013, 20:07 | posts: 23,677 | Location: NZ

Quote:
Originally Posted by sykozis View Post
I decided to give this a try. I setup a 512mb read "cache" for my data drive. Copied files ranging from 700mb to 4.5GB. On the 700mb files, it was roughly 1-2 sec faster. On the 4.5GB file, it was noticeable slower. I retested the 4.5GB file with a 1024mb read "cache" with no improvement. Since I tend to transfer several GB of data from drive to drive on a regular basis...this software appears to be pretty useless for me. Since most harddrive benchmarks are synthetic, their results are meaningless. According to the data transfer progress window, the data transfer rate was higher with this software.....but the actual data transfer TIME was noticeably slower with my 4.5GB file (which is actually the full OpenSuSE x64 disc ISO).
Yeah to see an improvement the cache must be bigger than the file you're transferring, because once the cache fills up the data has to be written to disk - just like a normal drive.
This would defeat the point of having fancycache. Fwiw write speeds are not increased. It looks that way in benches because the data is copied to ram first.
However it still must be written to disk so the file transfer takes longer than without Fancy Cache.

I'm not trying to rain on anyone's parade here but it's really an outdated concept that has been made redundant by SSD's and 64bit Windows.
Even on X86 the "invisible ram" thing is a bit misleading due to AWE/PAE.
It's silly to purchase 8-16GB of ram for cache when Bob Smith could buy an SSD for around the same price.

And some people pay waaaaaay to much attention to benchmark scores......

Last edited by Pill Monster; 08-24-2013 at 23:16.
   
Reply With Quote
Old
  (#18)
sykozis
Ancient Guru
 
sykozis's Avatar
 
Videocard: eVGA GTX660SC SLI
Processor: Core i7 2600K
Mainboard: ASRock Z77 Extreme4
Memory: 8gb G.Skill DDR3-1866
Soundcard: Creative Recon3D PCIe
PSU: SeaSonic M12II 620 Bronze
Default 08-24-2013, 23:09 | posts: 16,075 | Location: US East Coast

Quote:
Originally Posted by Pill Monster View Post
Yeah to see an improvement the cache must be bigger than the file you're transferring, because once the cache fills up the data has to be written to disk - just like a normal drive.
This would defeat the point of having fancycache.
Fwiw write speeds are not increased. It looks that way in benches because the data is copied to ram first. However it still has to be written to disk so in fact the file transfer takes longer than without Fancy Cache.

I'm not trying to rain on anyone's parade here but it's really an outdated concept that has been made redundant by SSD's and 64bit Windows. Even on X86 the "invisible ram" thing is a bit misleading due to AWE/PAE.
It's silly to purchase 8-16GB of ram for cache when Bob Smith could buy an SSD for around the same price.

And some people pay waaaaaay to much attention to benchmark scores......
So, basically, what you're saying is.... the increase in file copy time using this software should be expected because I'm seeing the effect of what is actually happening.

There's a reason I used file copies to test this software. Most benchmark software is synthetic and doesn't show "real-world" results. Synthetic benchmarks are easy to fool whereas using "real-world" scenarios....you get real world results. It's sorta like running 3DMark11 to determine how your system will perform in Battlefield 3....instead of using Battlefield 3 to determine how your system will perform in Battlefield 3....


   
Reply With Quote
Old
  (#19)
Pill Monster
Ancient Guru
 
Pill Monster's Avatar
 
Videocard: 7950 Vapor-X 1175/1550
Processor: AMD FX-8320 @4.8
Mainboard: ASUS Sabertooth 990FX R2
Memory: 8GB Kingston HyperX 2400
Soundcard: Audigy 2 Platinum Ex 5.1
PSU: AcBel M8 750
Default 08-25-2013, 00:10 | posts: 23,677 | Location: NZ

Quote:
Originally Posted by sykozis View Post
So, basically, what you're saying is.... the increase in file copy time using this software should be expected because I'm seeing the effect of what is actually happening.

There's a reason I used file copies to test this software. Most benchmark software is synthetic and doesn't show "real-world" results. Synthetic benchmarks are easy to fool whereas using "real-world" scenarios....you get real world results. It's sorta like running 3DMark11 to determine how your system will perform in Battlefield 3....instead of using Battlefield 3 to determine how your system will perform in Battlefield 3....
Yeah because you were writing 4.5GB to a disk with only 512Mb of cache. For the sake of simplicity your 512MB cache will now be called RAM:

RAM simply acts as a buffer, so data is first copied to the 512MB of RAM u reserved, while simultaneously writing from RAM to the HDD.
But data is copied to RAM faster than it can be written from RAM to the HDD so it takes only a few seconds for the 512MB RAM to fill up.
Once that happens all data in RAM must be written to the HDD (buffer flushing), before new data can be copied to RAM. That's why your transfer times seemed slow, because they were. lol. Try again using an 8GB cache and you'll see lightning fast (reported) transfer times.

So yeah the guys above who had huge caches will see very high benchmark numbers because as far as CrystalDisk is concerned, the transfer is complete as soon as the file is copied to RAM, not when it's written to disk.


(Is this post really as long winded as it seems? lol)

On a side note disabling Write Cache Buffer Flushing can improve HDD write speeds in Windows, however I don't recommend doing this on the system drive.

Last edited by Pill Monster; 09-06-2013 at 20:13.
   
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump



Powered by vBulletin®
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
vBulletin Skin developed by: vBStyles.com
Copyright (c) 1995-2014, All Rights Reserved. The Guru of 3D, the Hardware Guru, and 3D Guru are trademarks owned by Hilbert Hagedoorn.