What gives with CrystalDiskMark? -- odd results, or won't run

BonzaiDuck

Lifer
Jun 30, 2004
16,122
1,738
126
I'm trying to remember when I downloaded the CrystalDiskMark benchtest program: My download may be at least two years old.

I had it installed on my "flagship" (sig-rig). Recently, I wanted to use the bench-results to compare two different RAM-caching programs on two different machines (other than the sig): PrimoCache and SuperCache.

On the last of the machines where CDM is installed, it seems to give reliable results. On the other system, I'm getting "f***y" results -- like a 4K read test showing 0.35 MB/s, where I should really expect more than 100 times that figure.

So just now, I tried to run it on my sig-rig, and it is completely inoperable. It won't run -- not from a shortcut, and not from the Start-Menu "Programs."

Maybe someone could point me toward a better shareware program for benching my storage systems.
 

R0H1T

Platinum Member
Jan 12, 2013
2,582
163
106
I'm trying to remember when I downloaded the CrystalDiskMark benchtest program: My download may be at least two years old.

I had it installed on my "flagship" (sig-rig). Recently, I wanted to use the bench-results to compare two different RAM-caching programs on two different machines (other than the sig): PrimoCache and SuperCache.

On the last of the machines where CDM is installed, it seems to give reliable results. On the other system, I'm getting "f***y" results -- like a 4K read test showing 0.35 MB/s, where I should really expect more than 100 times that figure.

So just now, I tried to run it on my sig-rig, and it is completely inoperable. It won't run -- not from a shortcut, and not from the Start-Menu "Programs."

Maybe someone could point me toward a better shareware program for benching my storage systems.
OR you could try the latest version & see how that runs ~

http://www.softpedia.com/get/PORTABLE-SOFTWARE/System/System-Info/Portable-CrystalDiskMark.shtml
 

BonzaiDuck

Lifer
Jun 30, 2004
16,122
1,738
126
Just to clarify.

I refurbed an old laptop (C2D T8300) with 8GB of RAM and (tentatively) a Crucial MX100 SSD. The controller, of course, was SATA-II and AHCI-compliant. Since the SSD isn't Samsung, I wanted to get an idea of how the lappie would be "snappy" using a RAM-cache program.

Typically, the SSD (without caching) would show 4K read results in the 30 to 50 MB/s range. With a cache using 3GB of RAM, the score increases to around 150.

So one might also have optimistic expectations for a simple WD Blue HDD on an Intel SATA-II controller. I was stunned to find the 4K results with "caching" at well below a MB/s.

Now I've done a web-search, discovering that HDD 4K bench results with an SATA-III controller aren't much better.

As for the programs, I mentioned my observations in another thread. As much as I'd like to favor SuperSpeed's SuperCache because their HQ is in Massaschusetts, and they are MS Certified Partner -- I have to recommend Primo-Cache from Romex Software (Shanghai).

Both programs deploy block-level caching. Both allow for setting the cache-size to any number (although Primo's drop-down menu only presents 512MB and 1GB increment options for the lazy mouse-click).

For the price of one PC license from SuperSpeed, you get a 3-PC license from Romex. Primo-Cache shows much better bench results. Primo allows for write-caching on a boot drive, while SuperCache doesn't. And the free trial period from Romex is almost outrageous: you can test the software for 90 days before paying for it. SuperSpeed only gives you 14 days.
 
Last edited:

R0H1T

Platinum Member
Jan 12, 2013
2,582
163
106
Just to clarify.

I refurbed an old laptop (C2D T8300) with 8GB of RAM and (tentatively) a Crucial MX100 SSD. The controller, of course, was SATA-II and AHCI-compliant. Since the SSD isn't Samsung, I wanted to get an idea of how the lappie would be "snappy" using a RAM-cache program.

Typically, the SSD (without caching) would show 4K read results in the 30 to 50 MB/s range. With a cache using 3GB of RAM, the score increases to around 150.
So what are your observations without taking into account the synthetic benchmarks ? I personally feel that an SSD doesn't really need a RAM cache to speed up the system's performance, in fact it's advisable to not use a program such as Primocache if the SSD is good enough since the performance difference will probably be minor in most cases. The moment your read/write data spills out of the RAM cache you'll see a major performance hit & as such one needs a fairly decent amount of RAM, say 16GB, to make the program real useful. Also the IMC/RAM/CPU & the OS' system process will be continuously pegged making the system slightly unresponsive in such cases where the read/write data won't fit into the RAM cache. This is based on my personal experience using Primocache on a laptop with 8GB ram & a TB of 5400rpm HDD but I guess you can shed better info wrt your SSD.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,122
1,738
126
So what are your observations without taking into account the synthetic benchmarks ? I personally feel that an SSD doesn't really need a RAM cache to speed up the system's performance, in fact it's advisable to not use a program such as Primocache if the SSD is good enough since the performance difference will probably be minor in most cases. The moment your read/write data spills out of the RAM cache you'll see a major performance hit & as such one needs a fairly decent amount of RAM, say 16GB, to make the program real useful. Also the IMC/RAM/CPU & the OS' system process will be continuously pegged making the system slightly unresponsive in such cases where the read/write data won't fit into the RAM cache. This is based on my personal experience using Primocache on a laptop with 8GB ram & a TB of 5400rpm HDD but I guess you can shed better info wrt your SSD.

Perhaps all enthusiast experiments are best preceded by the most detailed rational calculation. In the case of this old [Penryn C2D] laptop, I wanted to see how I could make it "responsive" and serviceable.

So far, I haven't spent a dime on a caching program -- especially owing to the extended trial period on PrimoCache. The problem with both the laptop and the desktop system (another C2D configuration) -- they both use controllers that meet the SATA-II standard with AHCI. This might not restrict the performance of an SATA-III HDD, but most certainly affects the performance of an SATA-III SSD. Without caching, the laptop with SSD still offers significantly better performance than with the WD 2.5" HDD which came with it two months ago. But it is less than half the performance one would expect under SATA-III.

With caching, and regardless which caching program is used, performance and responsiveness for mainstream desktop software is again improved -- less noticeably than the simple comparison of benchmark results, even so.

I suppose the real issue boils down to the meaning attached to the benchmark results.

As for the laptop "investment," I can always replace the SSD with the WD "blue" HDD, so it is less a case of frivolous spending than a possible suboptimal use of parts.

If I could keep my curiosity from influencing my purchasing habits, I suppose I could save more money. But I'm not the only "cat" in these forums -- I'm sure! :D

AFTERTHOUGHT: As rule of thumb, I'm adverse to adding "complexity" to my systems -- more so now than in the past. PrimoCache has this "L2" feature such that you could use a cheap <=60GB SSD for a persistent cache of an HDD. I'm pretty sure I can't fit two 2.5" drives (SSD and HDD) in the laptop; I couldn't find a second SATA+power plug inside the lappie. I MIGHT try it with the old desktop I mentioned.

I'll leave off here to observe that the lappie was a relative slug with only 2GB of SO-DIMM, and no matter what I do with the currently loaded software, it seems unlikely that I would use 4GB after opening a whole pile of programs. So the only benefit I can get with 8GB of SO-DIMM derives from use of a caching program. Howsoever my observations or perceptions are inaccurate, partial imagination or profoundly real, this is about as good as it gets for a surplus laptop I bought for $250 which once retailed for $2,200. As a "new toy" to play with, it is an antidote for the temptation to spend $1,000 on a new laptop with current-gen processor, controller and 8GB of RAM.

Romex offers a 3-PC license for approximately 2x the single-seat version, and I still have 70 days to "think about it." So far, it looks as though I could use the more expensive license version.

But there are obviously limits to resurrecting old technology, and every increment of improvement costs more.