SB 2500K Enable Hyper-threading?

I haven't seen any reviews or news stories about the upgrade card program being available on the 2500K.
 
gunderwood said:
Not if they laser cut the die, which I'm almost positive they do. Same with the cache.
Click to expand...
HT is entrenched throughout the execution pipeline. It is not something that could be "laser cut". :p

It certainly is a feature that Intel can lock out at will. No one has broken even the oldest Intel locks on these types of features, going all the way back to the multiplier locks introduced in the PII days (or HT lock on P4 > 7 years ago).
 
pxc said:
HT is entrenched throughout the execution pipeline. It is not something that could be "laser cut". :p

It certainly is a feature that Intel can lock out at will. No one has broken even the oldest Intel locks on these types of features, going all the way back to the multiplier locks introduced in the PII days (or HT lock on P4 > 7 years ago).
Click to expand...

The laser cutting of ICs is nothing new and it doesn't matter how entrenched it is. Single transistors can be isolated. Specifically, they wouldn't "cut it out" (don't be ridiculous) they would trim the power and/or clock interconnects so that HT is never even on. The rest of the circuit is designed to be connected or isolated with no ill effects. It isn't hard.

If fact we know Nvidia does this with their GeForce/Quadro cards because the SW "lockout" was too easily hacked. After production they physically alter the die so that you can not unlock a GeForce into a Quadro. They also cut out whole execution engines. Again, not literally in the sense that the chip has a hole, but they do cut the interconnects.

http://www.semitracks.com/index.php...die-level-analysis/deprocessing/laser-cutting

Laser cuts can be made for many different reasons and in many different types of materials. Lasers have been used for two purposes: selective layer removal and interconnect isolation.
Click to expand...

The program the OP references was a test program of sorts. AFAIK, it wasn't offered for very long and it was only on very low end processors. The fact that Intels "lock" hasn't been broken yet would suggest it is using physical die alternations to prohibit doing what the OP suggested. SW locks don't usually last very long. We simply don't know. However, I should note that when that program debuted there was a lot of shock because up until then all information indicated they physically made such a thing impossible. The program was likely a test run to try out a different business model, nothing more.

We don't know what they did to the 2500k.
 
Last edited:
gunderwood said:
The laser cutting of ICs is nothing new and it doesn't matter how entrenched it is.
Click to expand...
LOL, yes it does. HyperThreading isn't a separate block on the CPU, where "laser cutting" might be useful. Intel has much better ways of controlling features on the CPU, including fuse based isolation/configuration and a writable configuration block (AMD does this too). Some features appear to be less permanently locked and the G6951 is an example where enabling HT and additional cache is software unlockable. Fuses are used in other parts of the CPU, for example to disable bad cache blocks (there are redundant blocks available for this purpose) and is done during testing.

I can obviously tell you've heard about laser cutting, but the toolbox contains more than a hammer. :p

HT, at least on Lynnfield, does not seem to be permanently fused off in configuration. It is more likely written as a configuration value in the writable area, which is used during power on to configure the CPU. There is no reason to believe Intel changed it in SB. The mystery is why it's so hard to reconfigure these CPUs. Strong encryption to enable the writable area is most likely the cause.
 
pxc said:
LOL, yes it does. HyperThreading isn't a separate block on the CPU, where "laser cutting" might be useful. Intel has much better ways of controlling features on the CPU, including fuse based isolation/configuration and a writable configuration block (AMD does this too). Some features appear to be less permanently locked and the G6951 is an example where enabling HT and additional cache is software unlockable. Fuses are used in other parts of the CPU, for example to disable bad cache blocks (there are redundant blocks available for this purpose) and is done during testing.

I can obviously tell you've heard about laser cutting, but the toolbox contains more than a hammer. :p

HT, at least on Lynnfield, does not seem to be permanently fused off in configuration. It is more likely written as a configuration value in the writable area, which is used during power on to configure the CPU. There is no reason to believe Intel changed it in SB. The mystery is why it's so hard to reconfigure these CPUs. Strong encryption to enable the writable area is most likely the cause.
Click to expand...
Yes, there are other ways to do the same thing. I'll admit my OP was poorly worded; I should have said I think it is permanently disabled rather than specifying a specific method. I'd love to see your proof that it isn't permanently fused off, cut, <insert favorite way to disable a portion of an IC here> or are you just guessing?

As a side note, if they can fuse off a circuit they can also isolate that circuit with a cut interconnect(s). The point of all such physical alterations is to remove power and/or clock to a portion of a circuit so that it can't work.
 
Most fuses aren't going to physically cut off a clock or power connection; the value of the fuse will get mapped to the chip's register space coming out of reset.
 
Too bad we can't do the "pencil trick" like on some old AMD CPUs :)

I really wanted the Hyperthreading of the 2600K, but, couldn't justify the more than $100 price difference so I ended up with the 2500K.

If I could pay $50 a year from now to "upgrade" my chip, I probably would (prices would come down by then).
 
Hyperthreading is more for encoding. I'm running a 2600k highly overclocked with 4/c 4/t meaning hyperthreading disabled and games actually run alot better. Windows is damn near the same, as is most tasking. It's mainly in encoding type stuff you do notice a decent difference.
 
You must log in or register to reply here.
Back
Top