New Turbo Memory Tests Show it in New Light

AnandTech's original article on Santa Rosa and the associated Turbo Memory technology left a lot to be desired. It appeared as though there were absolutely 0 benefits, and in fact performance was degraded in certain cases.

Today, AnandTech has written a new article on Turbo Memory after speaking with Intel about the matter. They've changed up some of their benchmarks to reflect how Turbo Memory can affect performance.

http://www.anandtech.com/mobile/showdoc.aspx?i=3009

Needless to say, it seems like the biggest advantage here is battery life (which in itself might be worth the $30 or so it's going for after-market). Some of the transfer rate tests show that it could potentially have an impact on performance as well, but will depend on the application. It would also seem like the NAND used is somewhat slow - write operations to a 7200RPM disk being buffered through the Turbo Memory slows it down quite a bit.

 

This thing is history when SSD goes mass.

 

Which won't be for several years. I think it has some life yet.

 

I think we are all waiting for Robson 2.0

 

I thought it was just driver/software support issues that kept it from working the way it was supposed to. What happened to Intel's demonstrations of starting up notebooks twice as fast?

 

Curious how much it would benefit a 5400 rpm drive...

 

If what you care about is primarily performance, then between adding 1GB of RAM or adding 1GB of Robson, adding 1GB of RAM is the best bet. It requires no special drivers and always works. Here is what Anandtech has to say about this:

And:

No Turbo Memory for me.

 

Looks like all TM does is boost performance and battery life when you're doing next to nothing but when you're pushing your system it does nothing or actually slows it down. I'll go with the extra gig of RAM over the TM option thanks.

 

Even after Intel gets their word in the benchmarks are still bad, as anticipated.

 

What really annoys me is that some really good brainpower was spent on a half-assed solution instead of less buzzword-friendly but better understood technologies. Good grief, I'm flashing back to the segmented memory of the 8086. Yes, it was cheap. Yes, it allowed software to limp past the limited memory addressing capabilities of the 8086. But it was an abomination from hell! Many an engineer cursed Intel for that one.

 

Too bad it cant just show up as a mass storage device that any system can use, so older non-santa rosa could benefit from it. Its not like its physically incompatible with older systems, its just a miniPCI-e card.

 

You just want to be able to jam it in your T60.

Sniff sniff me too.

 

The thing is, Turbo Memory is not designed to replace RAM. In fact, it is supposed to complement the cache built into the hard drive, not act as a buffer for the CPU (ie RAM). Unfortunately, it doesn't seem like it's doing much for hard drive performance either.

 

The fact that Turbo Memory is not designed to replace RAM is irrelevant. The important point is that RAM can do almost everything Turbo Memory can. The only two things RAM cannot do are persist across reboots and use as little power as Turbo Memory.

 

I wonder how it would work with the Seagate Hybrid drives? Anyone have a link to a test between Seagates Hybrid drive and its 7200.2 drive? Thanks.