Are solid state hard drives faster?

t
y

Thanks! Helpful reply. I guess I read this as: try to avoid pauses, wherever they may occur. Sounds like good advice!

Reply to
mpm
Loading thread data ...

You can only go as fast as the IO SATA or what ever you have, can go.

Many current HD's can out do this limit how ever, it's only if the data that is being read is contiguous for the drive so that there is no pause in data.. A well maintain HD can operate rather good..

The more important think is, life time of writing and reading and the sudden move on access which is faster than a HD.. If your HD goes to rest, the SSD would have the data at the port waiting for your system to suck it up before your HD was up and spinning.

Its those momentary moments is when you really notice the difference.

Jamie

Reply to
Jamie

A RAM drive is a virtual hard drive contained in RAM. Much faster than SSDs or hard drives, but all data gets lost at power off.

--
- Mike
Reply to
Mike Warren

at

ly

.

As opposed to what? Soft and thin, like a hydrocephalic's?

Good. I assume yours is too.

I do know that a pet peeve of mine is that "solid state" doesn't mean "no moving parts". Because if it did, vacuum tubes are solid state. Are they, arse pirate?

Reply to
a7yvm109gf5d1

Everyday use of HP DL360 and 380 Proliant series.

2U configuration on the latter. That model has a whole RAID string of drives across the face of it. All hot-swappable.

All 150GB 1.5" 15k rpm SAS (Serial Attached SCSI) Ultra SCSI drives.

You lose... again... as usual.

formatting link

formatting link

Reply to
TheQuickBrownFox

Look for the reviews and look for failure mode stats, if any even exist.

Reply to
Chieftain of the Carpet Crawle

You seem to have problems accepting that the industry adopts names to promote a product or a feature of a product that sets it apart from another, thus a new family or sub division of an existing product.

In this case, we refer to hard drives, which have for decades been known as an analog medium, because it is analog at the read/write transducer itself. Behind that transducer, for all a chump like you knows, it gets digitized immediately. From then on, it is "solid state", BUT it is never, nor has it ever been referred to that way.

ONLY NOW, after we rid ourselves of the both mechanically in motion, AND analog storage/retrieval means, and move ourselves into the purely digital storage medium domain, it also happens to be a non-mecahnical device. Thus, "solid state hard drive".

You do not get to concatenate the terms that are used into partial references and act as if someone other than yourself got it ALL WRONG.

Try again, UpChuck. You ain't real bright.

Reply to
TheGlimmerMan

That's because compiling code is CPU limited, not disk speed limited.

The intermediate files are cached in memory, don't even have to hit the disk.

Grant.

Reply to
Grant

:

so

else

ope.

d

s?

.

em,

Thus far only win7 has "trim", so it is the only OS officially SSD ready. Opensuse provides some tips for SSD use. I don't know about lesser versions of Linux, but fedora probably has tips too. [Suse and Redhat are the enterprise flavors of linux.]

formatting link

On my win7 box with SSD, I turned off the virtual memory (killed the page file). That prevents useless writes. I also don't use indexing since the drive can be searched so quickly.

At this point, I'd say it is Intel SSD or don't bother. Just poke around the forums and you will find few complaints regarding Intel SSD in correctly configured systems (i.e. Win 7 where the user made sure trim was actually enabled).

Since you don't explain your requirements very well, it is hard to tell if more ram would be a better solution. I've never found a problem with too much ram. ;-)

Reply to
miso

Jamie > the speed of getting to the sector makes up for a lot.

Nymnonuts > No shit. That is why we pros use Nymnonuts > 1.5" SAS drives that spool at 15k rpm.

G > "we pros"?? LOL G > The only thing you are professional G > at is BS'ing, you autistic psychopath.

formatting link

formatting link

Post the link to your game box photos again, Archie! Make sure to paint out your mommy's garage in the background.

Reply to
Greegor

It isn't a 'game box', you dumb fucktard.

Your IQ is about 20, at best. 100 is average, BTW.

Reply to
TheGlimmerMan

No. The servers in question are already at their limit of maximum physical memory installed and the database index cache barely fits in. Some of these databases are absolutely huge with multi-terabyte sizes the SSDs are being used to speed up repeated access of popular data.

Even the 6 man chess tablebases run to over 1TB of highly compressed files and work is underway now to generate the 7 men ones for interesting cases (which are often the largest files).

Of the most tricky 6 men databases KBNP vs KQ alone runs to nearly 13GB in its compressed form and expands to a lot more when in use. Most of them are typically 1-2GB but the important ones tend to be the largest.

Regards, Martin Brown

Reply to
Martin Brown

On a sunny day (Tue, 29 Mar 2011 22:15:47 -0500) it happened Jamie wrote in :

I had a discussion about FLASH in some other group, with somebody who I think is from Intel (he would not admit). But he warned me about the effect of cosmic rays on FLASH memory, and sort of suggested they were having huge problems. I was a bit sceptical about that until later I heard that Intel retracted some project or model because of that same reason. So, considering long term data integrity that maybe something to worry about. So far my memory sticks and cards seem to be OK though, you can dump the device into a file and years later read it back and compare, only had one failure, but I stopped testing for that.

Reply to
Jan Panteltje

With regards to wearing out SSD drives; Just for the hell of it I've been running a continuous write and verify on a 8GB compact flash configured as a disk drive since July of 2009. The compact flash is made by Transcend and marketed as 100X speed fixed-disk industrial (much slower than the SSD drives available today). As of today it has written 270TB to the drive and verified the data with no errors. I doubt most users will abuse their flash drives in this way, suggesting that one would have a much longer use under normal conditions. Also, most SSD disk drive manufacturers use "wear leveling" which prolong the life of the drive. See

formatting link
It is important to choose a drive that will have a lot of free space once your OS and apps are installed, which will keep the drive error free much longer.

Reply to
JW

Are you write/verifying changing data? Continuous write/verify same data is a NOP, and wont wear the memory cells. The problem is that erasing Flash involves pushing some electrons through an insulator, the insulator objects.

So if the cell data is not changed for a write, that write may be skipped by the controller as unnecessary, as part of the speed up optimisation.

Grant.

Reply to
Grant

I had to opportunity to ask a high energy physicist about cosmic rays and computer memory and he told me the cosmic rays are stopped by the roof, they don't get into the PC at all. Yet it's common to hear people discuss cosmic rays as something that can affect terrestrial computers. It is a problem for the satellite builders.

There used to be an issue with the semiconductor packaging materials being very slightly radioactive, but that was discovered, countered decades ago.

Another problem with Flash now is the multi-level cell technology, expecting a cell to keep data 1 or 0 for a long time one thing, but running 4, 8 or 16 levels per cell is asking it to remember an analog voltage for a long time. The faster, cheaper flash memories coming out a year or so ago were 8 level (3 bits/cell). No idea what's being produced now, this year.

Grant.

Reply to
Grant

Running this diagnostic:

formatting link
According to the documentation:
formatting link

I am cycling between these patterns: · Default (Cyclic) · Sequential data pattern (0,1,2...255) · Random data with random seeking · High Low frequency data overwrite (10101 then 00001) · Butterfly seeking · Binary data pattern 1 (10101010) · Binary data pattern 2 (01010101) · Zeros data pattern (00000000) · Ones data pattern (11111111) · Random data pattern

I have it configured to test until the drive is full, then it delete the files and starts over. I think it's a pretty thorough test. (Now maybe some f****it who knows f*ck-all and shall remain nameless, but has the initials AW will stumble into the thread and accuse me of being a shill for Passmark.) :-)

FWIW, with Transcends Commercial grade 133X media, it failed at 21TB. With their 600X media, it failed at 53TB.

Reply to
JW

On a sunny day (Wed, 30 Mar 2011 21:18:59 +1100) it happened Grant wrote in :

I can assure you cosmic rays have no problem reaching the inside of a metal can deep in your basement if you have one. The radiation detector projects I have build trigger on those. You actually need a dual detector to detect the cosmic rays, for example

2 scintillation crystals and 2 PMTs, and then look for coincidences. Those very short pulses happen at almost the same time when a cosmic ray passes first through one, then through the second detector. You can also make a vapour chamber with some alcohol and a Peltier cooler, to make these things visible.
formatting link

Yes, may as well store the audio in analog form :-) LOL

Reply to
Jan Panteltje

You are an electrical dumbass (must be). An electron CANNOT "pass through an insulator". Not without boring a hole through it. If no hole, it is NOT "an insulator". If it makes a hole, it is no longer "an insulator". The very definition of "insulator" is such that NO electron will EVER "pass through it". Bad term choice by either you or the industry.

I do not think it "looks" to see if data ready to be or instructed to be written is already present. No such comparison is made. You tell it to write data, it writes the data. Anything else is a "drive failure" mode.

Despite your intelligent responses thus far, this one is in err. It goes against the very file system protocol(s). The "drive" device is not allowed to refuse or "skip" a write and NONE perform this magical comparison you mention.

Reply to
Chieftain of the Carpet Crawle

The difference is that if you just use OS disk cache, the OS will still want to write the data to the hard drive now and then, which will take time. With a real ramdisk, there are no hard drive accesses at all.

John

Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.