eBoostr and Superfetch

[chooky]

Just finished a round of testing eBoostr with several applications and investigation into performance with MS Superfetch turned off.

For these tests I used eBoostr 4 beta build 535 with a 2GB RAM cache on Windows 7 x64 RTM with 2 x Vertex 30GB RAID 0.

I used Passmark Apptimer to time application loading.



The results show that if you have an SSD it is for the most part better to have Superfetch turned off although some programs will suffer.

Adding an eboostr RAM cache decreased first load times by a good margin. Subsequent loads did not improve much.

Turning off Superfetch gave a further decrease in first and subsequent load times.

My PC does feel snappier because eBoostr caches a lot of system files as well as programs.

Boot and shutdown times did not change by more than a second.

Other benchmarks like Passmark Performance Test or PCMark Vantage did not show any difference.

More setup tips:
It is important not to use a USB cache only memory.

It is also important to go to Edit -> Show cache content in the control panel and right click on entries for programs you don't use often, as well as C:\Windows\System32\catroot, and exclude them by path. This gives more cache space for programs you do want to speed up. Then rebuild the cache.

Check with the eBoostr Cache Statistics program in your start menu to make sure the programs you are testing are cached nearly 100%, right clicking the item and increasing priority if they are not, or they won't be faster.

To get these sorts of SSD performance increases from a US$40 software solution is pretty amazing.

[HDCHOPPER]

Thanks chooky for your efforts & the info

[cavd]

For these tests I used eBoostr 4 beta build 535 with a 2GB RAM cache on Windows 7 x64 RTM with 2 x Vertex 30GB RAID 0.

I'd be interested to know total system RAM, which of the "standard performance tweaks" were in place (pagefile off, indexing off, etc), and what the numbers look like on a un-overclocked configuration...

[chooky]

Total RAM 4GB, indexing is off, 1GB fixed Pagefile only needed if you run large programs like games. The important aspect of your hardware performance is the Disk speed vs Memory Speed ratio. eBoostr provides a test for both.
These are my results:
---------------------------
eBoostr : Control Panel
---------------------------
Direct access speed: 228.011597 MB/s

Cache enabled access speed: 892.565735 MB/s

Ratio: 3.914563

So with a 1:4 ratio you should get the same results. If you are running a HDD you ratio will be better therefore results will be better.

[cavd]

I think you've slightly missed the point of my questions, and I don't wish to sound like I'm disparaging your efforts in producing the statistics... however...

1GB fixed Pagefile only needed if you run large programs like games.

Does this mean that the tests were done with or without a pagefile?

The important aspect of your hardware performance is the Disk speed vs Memory Speed ratio.

Of course this true, but your figures will be somewhat skewed as you're apparently running a 2.83Ghz CPU at 4Ghz -impressive in it's own right but likely to overstate the benefit of RAM-based caching at best; and given the potential combinations of CPU multiplier, FSB, PCIe and RAM frequencies, be completely meaningless at worst.

Again, don't get me wrong, I think you've undertaken a laudable exercise... I just feel that doing the same metrics at stock speeds would give us a much better idea of the benefits... without going to the effort of doing the tests ourselves

[chooky]

The tests were done with the pagefile active. When I can afford it I will purchase another 4GB so I can have a 4GB cache and no pagefile.

I can see you point and agree that my results may be better than what you would get from a PC running at stock.

Newer hardware with better memory (DDR III) or faster memory mine is DDR II 1066MHz and or better memory controllers like the P55 or X58 chipsets will perform better at stock than my PC at stock.

What is common with most PC's is that they have free RAM that is not used most of the time. This ram is nearly always faster than storage speed. The software is free to try. If beta software sounds risky the version 3 has a trial period. I strongly suggest trying this software to see if it is a worthwhile upgrade on your own particular hardware.

[Vorenus]

Great tests chooky.
And I've learnt about AppTimer which is an app I didn't know. Used to write my owns everytime for this purpose before.

The apps benchmarked here are relatively light, it would be interesting to see what the gains are with more massive applications... like Photoshop or even better Visual Studio 2008 if you have them?

[chooky]

This programmer is using eBoostr to reduce compile times by over 50% in Visual Studio.
http://dotnet.agilekiwi.com/blog/200...with-slow.html
http://dotnet.agilekiwi.com/blog/200...str-works.html

[Pyrolistical]

It doesn't cache writes, which is the slow bit in SSDs.

This is a good idea, but a poor implementation. You don't want to consume the ram as cache, you just want to add more cache.

This is a great opportunity for some company to create a battery backed SATA adapter that has 3-6 slots of DDR3. This product will cache writes and if the power is pulled, the cache will be flushed to a SSD. These things will improve the performance of both SSD and HDD. Its just that on read cache miss SSD still has a huge advantage with its low latency.

Cache on the SSD is to hide performance issues at the NAND level, this product will hide performance issues on the SATA level.

[fqvarfort]

This programmer is using eBoostr to reduce compile times by over 50% in Visual Studio.
http://dotnet.agilekiwi.com/blog/200...with-slow.html
http://dotnet.agilekiwi.com/blog/200...str-works.html

... when using a 5,400 rpm HDD. If he chose to install a "2nd generation"/decent SSD I am sure the difference would not be as noticeable, if not negligible.

[chooky]

fqvarfort, If an application is mostly cached in RAM it makes no difference how fast the storage is. This is the great thing about eBoostr you don't have to buy an expensive SDD to get better than SSD application opening performance you just need enough RAM and eBoostr.