Opinion on current ThinkPad trends for a feature story

[dr_st]

Drivers for old thinkpads isn't as big of an issue as the model # lookup and HMM. IMO, most of the drivers are old and only work with outdated and insecure operating systems.

It depends how far back you go. Certainly some of the 4x/3x and 6x series machines have drivers even for Vista and Win7. Not to mention that if Lenovo really implements the ridiculous "rolling 7-year window" policy, then list of "obsolete" models will quickly expand to include even some relatively modern machines, which are very capable running the latest versions of Windows.

[Ibthink]

You would likely be much better served by installing Linux, which still supports much of the hardware out of the box (puppy linux, debian, etc.) and with many apps to choose from, including capable browsers.

Indeed - whats the point of using an outdated Windows like XP that is insecure and slow...

Getting rid of the model # lookup and HMM is a terrible idea. With no model lookup, it makes diagnosing and buying older thinkpads more difficult, and with no HMM, you risk breaking/damaging something or going around in circles trying to figure out how to install or change out a part.

Regarding the model number lookup, I hope they will update the new site to include the model # of the middle-old ThinkPads (T40-T400), because this situation is very bad if you want to look up the specs of a older ThinkPad system - my Z61t and my R50e are also not recognized.

Regarding the HMM, I hope they will set up a sort of "legacy site" - German forum member Mornsgrans noted that Mark_Lenovo needs all the criticism in the Lenovo forum that he can get (as long as it is polite) so that he has a stronger internal position on this matter - because to set up such a site, he needs a permission from the higher managment.

[dr_st]

Regarding the model number lookup, I hope they will update the new site to include the model # of the middle-old ThinkPads (T40-T400), because this situation is very bad if you want to look up the specs of a older ThinkPad system - my Z61t and my R50e are also not recognized.

Regarding the HMM, I hope they will set up a sort of "legacy site" - German forum member Mornsgrans noted that Mark_Lenovo needs all the criticism in the Lenovo forum that he can get (as long as it is polite) so that he has a stronger internal position on this matter - because to set up such a site, he needs a permission from the higher managment.

But why is it important to recognize R50e and not the T23? What cutoff date do you find reasonable and why is it more reasonable than their current arbitrary one?

I fully agree on the legacy site, except instead of wasting more time and resources on creating something new and crippled, they should just keep the old site as legacy only, forever, without any cuts and without any future updates. That's what I've been proposing from the start, and I think it's very reasonable.

[brchan]

The oldest HMMs Lenovo still has up seems to be for the T4x and X3x series. Anything older seems to have been removed.

At this point, I doubt Lenovo will listen to any criticism. But I hope I am wrong. They have become too obsessed with money and has turned a blind eye to its customers (a fate all too common for many big businesses).

[pianowizard]

Indeed - whats the point of using an outdated Windows like XP that is insecure and slow...

Nearly a quarter of all computers still use Windows XP: http://www.netmarketshare.com/operating ... pcustomd=0 .

For Windows 2000 and 98, the market shares are 0.03% and 0.01% respectively, which still amount to hundreds of thousands if not more.

At this point, I doubt Lenovo will listen to any criticism.

Especially since there are people who blindly defend Lenovo as if it could do no wrong, assuring people at Lenovo that there's no need to change anything.

[dr_st]

The oldest HMMs Lenovo still has up seems to be for the T4x and X3x series. Anything older seems to have been removed.

Everything is still here:
http://support1.lenovo.com/en_US/detail ... D=NDEX-HMM

[600X]

Nearly a quarter of all computers still use Windows XP: http://www.netmarketshare.com/operating ... pcustomd=0 .

Never mind those, what about the Windows 7 users? I still need those Win 2000 and XP drivers to get Win 7 running on the old machines.

[sriram00]

Hello everybody,

I started this thread for your opinions on a story I was working on. It's now online.

http://thewireless.co.nz/themes/impact/ ... le-replace

It has inputs from some forum members and even from David Hill. Got my queries answered by him over email. Could not use that entire interview in my story.

Thank you all for your feedback.

[pkiff]

Good article. Nice work.

Phil.

[Temetka]

Great article.

I've been a member here for years and I learned something new.

Ray works for Samsung.

Now I know who to pester with this one question: When are you guys going to update my N8013?!?

[rkawakami]

Sorry, but the group I work for is developing the technology called STT-RAM. We were originally founded as Grandis, Inc. before being bought out by Samsung a couple of years ago. STT-RAM replaces traditional electrical data storage (i.e., transistors and/or electrical charges) with magnetically-switched cells. In the simplest terms, it is a method of flipping (writing) a magnetic cell into one of two states (parallel and anti-parallel). It's considered by most people to be a "universal memory" in that it should be fast as static RAM, take up similar-sized silicon real estate like DRAM and be non-volatile like FLASH, without being saddled with FLASH's finite number of write/erase cycles. All of this being done using a very small amount of power. The one drawback of this technology however, is that it can be sensitive to strong magnetic fields.

My job is to write test programs that evaluate and characterize the performance of this type of memory. Unfortunately, our group has nothing to do with the semiconductor, hard drive or the PC end of Samsung's business. I do use a company-issued Samsung laptop (model escapes me at the moment) and have a Galaxy Note 8.0 at home but that's about the extent of my exposure to Samsung computers.

[dr_st]

I started this thread for your opinions on a story I was working on. It's now online.

http://thewireless.co.nz/themes/impact/ ... le-replace

Very good article. Great job!

[sriram00]

Thank you all.

Ray, your work is incredibly interesting. Pardon my asking, but can this technology be used for random access AND permanent storage? And is IBM also working on something like this?
http://www.engadget.com/2014/09/04/ibm- ... ard-drive/

Of course, being proprietary tech, we'd completely understand if you wouldn't want to discuss it at length.

But whatever your company is working on, great going and all the best.

[rkawakami]

It's not too proprietary to talk about as there are articles that have been published about STT-RAM, as well as IBM's racetrack memory design. The STT-RAM design that I'm currently working with is essentially encased inside a static RAM (SRAM) interface. This means that the device is able to uniquely and randomly address a single memory bit in a small-scale array (we're talking much less than 1Mbit for this engineering project). Give the device an address, provide some clock pulses to write or read and that's it. SRAM interface is much simpler than DRAM or FLASH but there's really nothing to prevent using those architectures when the technology is mature. The actual memory bits in the array are physically laid out very similar to a DRAM design where there's one transistor and one capacitor that make up the memory bit. With STT-RAM, the capacitor is replaced by a magnetic tunneling junction (MTJ). Where a DRAM senses the absence or presence of an electrical charge stored in the capacitor, STT-RAM looks at the resistance to current flow through the MTJ. The MTJ is structured to have a "fixed" magnetic layer and a "free" magnetic layer. By passing enough current through the MTJ, the free layer can be flipped between two states: parallel to the fixed layer or anti-parallel. By applying a much smaller current (e.g., 10x less than the write current), there will be some resistance to the current flow depending upon the state of the free layer. Parallel state is low resistance and anti-parallel state is somewhat higher resistance. When this difference is detected properly, this results in the basic binary states of a memory bit (low/0 or high/1).

Using magnetic bits to store data should be "permanent" in the respect that you don't need any power to maintain the data. This is where STT-RAM is seen to be equivalent to FLASH and much better than DRAM (which must constantly be refreshed even if not actively being used, drawing significant power when doing so) and SRAM (which does not require any refreshing, uses much less power than DRAM when idle but does need constant power to maintain data). Think of it like a hard disk drive without any moving parts. That said, there's still some things that can cause memory loss in an MTJ. As I mentioned earlier, an external magnetic field can cause bit changes if the device is not properly shielded. In fact, this is how I test the silicon wafers here at work. I have an electromagnet mounted above the wafer and by applying positive and negative voltages to it, I can "write" all of the bits in the memory array to the parallel (data=0) or anti-parallel (data=1) states.

FLASH has a finite number of write/erase cycles for each bit. Given the current design technology, it's typically measured in the order of 100,000 to less than 10,000 cycles. DRAM and SRAM usually don't wear out as long as there's a proper "burn-in" applied to them before leaving the factory. STT-RAM should be infinite or very close to it . It's still a developing technology that needs to be carefully looked at. MTJs can be damaged by forcing too much current through them; for example, using a write voltage pulse that is too high. Conversely, if the write pulse does not apply enough current, the MTJ may not switch states. There's also a phenomenon known as "backhopping". Simply defined, it's when you attempt to write the bit with the same data that's already stored there, it can flip to the opposite data state. The probability of this happening is fairly small but it does increase as the write voltage increases.

I was not familiar with IBM's racetrack technology but it does seem to use some of the basic properties of magnetic domain switching. Their memory design is more like a shift-register than a single bit storage cell. I know in the past that IBM has worked on STT-RAM technology with a couple of partners but I can't speak about what they are doing now.

[ajkula66]

Very good article. Great job!

Agreed thoroughly! Keep up the good work.

[Temetka]

Sorry, but the group I work for is developing the technology called STT-RAM. We were originally founded as Grandis, Inc. before being bought out by Samsung a couple of years ago. STT-RAM replaces traditional electrical data storage (i.e., transistors and/or electrical charges) with magnetically-switched cells. In the simplest terms, it is a method of flipping (writing) a magnetic cell into one of two states (parallel and anti-parallel). It's considered by most people to be a "universal memory" in that it should be fast as static RAM, take up similar-sized silicon real estate like DRAM and be non-volatile like FLASH, without being saddled with FLASH's finite number of write/erase cycles. All of this being done using a very small amount of power. The one drawback of this technology however, is that it can be sensitive to strong magnetic fields.

My job is to write test programs that evaluate and characterize the performance of this type of memory. Unfortunately, our group has nothing to do with the semiconductor, hard drive or the PC end of Samsung's business. I do use a company-issued Samsung laptop (model escapes me at the moment) and have a Galaxy Note 8.0 at home but that's about the extent of my exposure to Samsung computers.

Figures. Now I will never know when the N8013 is getting a SW update.

Awesome work on the RAM. I have an AS in EE and am pretty sure what you're doing blows my training out of the water. Got any non NDA schematics or whitepapers you can share? Sounds like great stuff. This is what I get for going into Network Security with a BS instead of something cooler like electronics or biotech. (But I do love Network Security though).

[rkawakami]

I don't have design documents to post on the current chip I'm testing, but here's a PowerPoint presentation (formatted as a .PDF) that one of our EVPs made before our merger with Samsung, outlining STT-RAM and Grandis' history up until several years ago:

http://nvmw.ucsd.edu/2010/documents/Dri ... xander.pdf

There's a lot of physics involved which I don't claim to understand, but with the direction given to me by the PhDs here, I've been able to create test programs to analyze the performance of this technology.

[Temetka]

I don't have design documents to post on the current chip I'm testing, but here's a PowerPoint presentation (formatted as a .PDF) that one of our EVPs made before our merger with Samsung, outlining STT-RAM and Grandis' history up until several years ago:

http://nvmw.ucsd.edu/2010/documents/Dri ... xander.pdf

There's a lot of physics involved which I don't claim to understand, but with the direction given to me by the PhDs here, I've been able to create test programs to analyze the performance of this technology.

Downloaded and placed on my never to be upgraded Note 10.1.

Thanks Ray, you rock.