Hello all,
While randomly surfing the web I found this great article about Intel's Turbo Memory. Apparently the authors spoke directly to an Intel engineer and managed to explain it with some detail but keeping it simple.
IMPORTANT: According to this article, you should NEVER mess with the settings in the Windows Vista Control Panel that deal with ReadyBoost or ReadyDrive. When you change them it becomes impossible to know if your Turbo Memory is in fact on or off!
It seems like an simple, unbiased, technically correct article. Check it out!
-
-
Hmm..that article says that you shouldn't mess with the checkboxes in the turbo memory console - *whoops* a bit late now
Now I have no idea if my turbo memory will actually be on or off!!!! -
Yeah, it's quite astonishing how something like that actually made into the open. I'll put a remark on the main post drawing attention to that.
By running some of the tests mentioned in the article you might be able to find out. Or maybe you can send an email to Intel about that... -
I scanned through the article, same as what people have concluded here, it is still better to have more physical RAM. If you have 2 GB or more RAM, the TurboMemory isn't that useful.
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3009&p=7
-
Turbo memory is a useless technology. It has a low data throughput of around 35mb/sec, which means that if you have a decent 160gb 5400rpm drive or faster, the turbo memory will do nothing for you.
As mentioned, the more memory the faster ones performance in Vista. The turbo memory module does not work as well as advertised. It is advertised as turbo because Intel knows it sucks, and they still want to be able to sell them to people who do not know how bad they are.
In the end, the turbo memory modules cause stability issues, bsod's and do not speed up the system at all.
K-TRON -
For Turbo Memory to be useful it needs to have data throughput of at least 80MB/sec and even still, more memory would actually do a better job for the same price or less. Turbo Memory 2 is around the corner with Montevina releasing in a couple months, and we'll see how much that actually improves performace, if at all. -
Not that I am trying to support ITM (see my sig where I mention that it sucks), but the 35mb/s of the ITM actually can be useful for things like random or nonsequential reads and writes. When you look at hard drive transfer rates, those are ideal rates for burst. However on the other hand to be useful for random read/writes, the data has to be located or cached into ITM, and with only 512mb/512mb for turbocache/readyboost, I'm not sure I'm confident it happens. I still leave mine in because I'm maxxed out in the RAM department and it is running 99.9% stably (I did get a BSOD related to ITM once before but that's about it). I've debated removing it but I cannot find that benchmarks or performance increases when I remove it. In fact I don't show performance increases or decreases with or without it when I benchmark (WEI, Sandra, rmma etc). I had a hunch before that it would actually decrease performance if installed because of my fast hard drive, fast memory (DDR2 667mhz at 4-4-4-12), and fast CPU but I haven't been able to either prove its usefulness nor disprove it.
Note: I found ITM to be much more stable after installing Vista SP1 - if not for that I likely would have removed it (as my previous posts seemed to indicate)removed per user -
You really don't get it. With all you have said you clearly don't get it. That is cool, if you are happy so be it.
-
If you can't prove it's usefulness while using it, or even notice a difference, I doubt it's actually even useful. -
-
Seriously, Intel really needs to apply this Turbo Memory system into a "Superfetch" caching algorithm and caching hardware mechanism that either uses actual memory DDR2/3 SODIMM modules or ultra fast NAND chips on miniPCIe 8x to cache program files capable of over 500MB/sec transfers(system memory is way higher than that), eliminating any kind of random access data required directly from the hard drive so that that data required by Windows OS would be nearly instant. Of course using an SSD would see less of a performance impact, but SSDs are still slow compared to this level of memory transfer speeds. I can't see why this won't work, it's not a hybrid HDD system as it would be tied with the Windows operating system on an BIOS/ACPI level. The only thing is that the algorithm would require CPU calculations from time to time, that could bog down the system, much like file indexing does in Vista.
-
I wonder when Intel will release a new turbo memory driver, the current ones are now almost 6 months old. It almost seems like Intel just kinda stopped supporting it, either that or they think the current ones are sufficiently stable. I suppose they are, at least for me, but I'm kind of anxious for a new driver. Maybe just my false hope that it will improve performance hahaha.
I guess they don't implement turbo memory as you mentioned above because the target price for the module (retail end user price) is about $50, and they don't want it to cost the same or more as a simple memory upgrade. I think they don't position it as a 'high end' component like we want it to be, only more of a band-aid approach to solving low memory issues in lower end machines. But that's just my take on it - i need to put on my flamesuit before posting opinions before I get blasted for it like powerpack did earlier even though my stance on ITM is on the side of disappointment as my sig has always pointed out -
Its a gimmick. Please dont get it.
-
Here's a thought:
HDTune took a look at the Intel Turbo Memory module and came up with lightning fast access times (0.1 ms) BUT 33% CPU usage...
I don't think using USB Readyboost uses that much CPU, so maybe that's better instead? -
CPU usage is nothing..
Readyboost might be faster but both are no match against real RAM. They are worlds apart -
I have 4 gigs of 667 ram, with 1.5-2gb usually free, and expresscard ssd readyboost gave me a pickup (previously sdcard readyboost), so the "I have enough ram" argument doesn't hold. Does anyone have real results, from 2008/SP1 (the anandtech article is old) with lots of ram and an ITM chip?
-
i conducted a passmark test a while ago. before using readyboost my laptop had a rating of 401.2, after applying readyboost my score improved by around 5 points to 406.2. This is not a fluke, it is the average of 3 tests. However a 5 point improvement may not be that noticable.
-
so the conclusion is ITM is not good enough ?
-
Jayayess1190 Waiting on Intel Cannonlake
Anyone who thinks turbo memory is dead, its not: Proof
-
lol! You resurrected a pretty old thread here!
In the incarnation that everyone here was talking about, yeah ITM was pretty useless. With the updated drivers it's gotten A LOT better though.
Before I sold it, I had put a 1GB turbo ram module in my Gateway M-6750. After I had done some hardware upgrading, I noticed that the module actually did give me some improvement in speed. I hardly ever run my systems off of battery, so I didn't see any improvement in battery life.
I just ordered a 4GB Intel Half Height Turbo Memory NVCPEHWR004G2 4GB PCIE Half Minicard for my new Gateway 7805u system. I'm figuring that with the new user pinning feature that the main applications I use (photoshop and lightroom) will be able to load faster.
Once it comes in I'll do some tests to see how it performs.
Investigating Intel Turbo Memory: is it useful or not?
Discussion in 'Hardware Components and Aftermarket Upgrades' started by HerrKaputt, Apr 23, 2008.