pretty damn impressive optimization guide

TweakTown's Ultimate Windows SSD Performance Installation Guide

check it out, not bad!

 

SE for each O/S install? Umm, no...

Disable cpu idling features for better scores? Maybe okay for a desktop system, but on a notebook forum?

Use old IRST drivers for their 'performance'? Lol... again, 'scores' are not performance in my stability/productivity oriented workflows.

Recommend RAID0 for performance in general (and I agree it is advisable for very specific workflows...)? No sense there either, because this performance advantage is never realized in most real world workloads. And saying that SSD's in RAID0 is safe because the author hasn't had an issue (yet) is like me saying that consuming fast food daily and in large quantities is safe because I'm still alive...

The tip about running 'winsat disk' was good though.

I too disabled my hibernation file with powercfg -h off right after a clean Windows install. And I used to disable the pagefile on my systems (with 16GB RAM or more), but with the latest fully updated Win8.1x64 Pro updates, I again revisited having the pagefile set to System Managed (not the 'auto' setting though) and have been experiencing a faster, overall user experience for the past month or so on all the systems I've tried it on (even with 32GB RAM or more).


The part about clean installing an update is appallingly stark. What is the benefit? What is the point?

I clean install Win8.1x64 Pro w/Update 1 from an original MS DVD (made bootable to a USB drive) and if/when a single install disk is released that includes any further updates, I will be using that one instead at that time.

But doing an update on a running system is not a 'clean install' - at least, not how it was (not) explained in the article linked.

The take away of the article to me is that SSD tests from TT are not applicable to real world usage; if that is how a system is setup/rigged to give the big numbers in the benchmarks.

Even a (desktop) workstation needs to sip power when on but not in use and turning off Win8.1 based computers in late 2014 is simply postponing the background maintenance they need to do, to simply be run when You want exclusive access to the computer (when you turn it on)...


Not bad? That's giving a little too much credit for a guide that would probably be better titled 'how to increase your BM scores' instead...

 

Beat me to it...

I too don't really see the point in trying to raise synthetic benchmark scores.

 

The guide's author got it all wrong, even according to his own mindset.

Why don't he just throw away all his sata SSD and get some PCIe SSD. Performance in the face, BOOM. :hi2:

 

sad thing is you can't raid pcie ssd yet and make it bootable, need mobo maker to get that started.

@tillerearth, SE to get people started, no need to do it everytime, and tbh I can tell speed difference between a drive with 25MB/s 4k random QD1 read compared to 35MB/s, although by doing the things he mentioned in article such as power tweaking and bios setting, dramatically reduces latency as well, means more snappy for SSD.

 

Also take a look at this. :thumbsup:

The SSD Optimization Guide Ultimate Windows 8 (And Win7) Edition | The SSD Review

 

This guide is probably more reliable then the one OP linked to..

 

lol more reliable.. SSD review is known to cover up sandforce trim issue with lies.
just to show how many ppl out to believe in that crap

 

I don't know about Sandforce drives and I couldn't give a crap about them but this article has helped me before so that's why I'm recommedning it..

 

SandForce-based drives have a long history of behind unreliable as well as underwhelming in terms of benchmarks and real-world use (SF uses performance tricks to work well with compressible data, but falls flat on its face with non-compressible data).

The only half-decent SF drives out there are Intel's SF-based drives, but only because they use their own firmware on it that works well.

 

funny thing is i have plenty of sandforce drive and none of them got issues. the non sandforce drive i have had problem and dying on me was crucial, samsung 840 pro, and 840 evo with recent issue that was just fixed. anyway, ssdreview is terrible, can't believe anyone actually read and believe their stuff

 

Well, as an overall trend, SF-based drives do fail more than non-SF drives (even a quick glance at Newegg reviews can show that). However it looks like you are lucky in that the drive works well for you. I'm also lucky in that my Intel 330 (SF) also still works (but I don't dare to put important data on it). However, the performance issues of SF are very real and are an issue in every single SF-based drive out there; this is especially painfully obvious for those working with a lot of nom-compressible data (anyone that uses encryption for a lot of their data, for example).

 

I thought the same thing about the problems with SandForce controllers until I read Dispelling misconceptions: Data reduction technology delivers only upsides to SSD performance | Seagate

 

I wouldn't trust an article from a manufacturer who uses the technology, but they only touched on concept, not the real issues. They didn't discuss speed throttling just to extend the life of the drive. The biggest issue regarding "DuraWrite", it prioritizes NAND life over performance. So even if you have a major workload and need peak performance for an extended period, it won't let you, because it wants to make sure the drive survives the warranty term...

But regarding the TT article, I think it's good. There's nothing wrong with an SE before a clean install. It will restore the SSD to best possible performance for the state the drive is in. And unless you do a clean install every month, it really isn't a big deal. Even power users do a clean install every 6-9 months, if that. Otherwise I agree (shocker eh) with tiller's statements.