T9300 CPU clock speed problem

[Cwrb]

I check my clock speed and it was 1190 instead of 2500.
Any ideas how to get it back to at least 2500?
I have a 7658-CTR.

[TuuS]

I have/had about 50 of these chips installed in various thinkpads and have no idea what your referring to regarding clock speed. These chips run at 2.5ghz, it's not adjustable, the multiplier is locked on the cpu and even if it wasn't, the bios doesn't support changing the setting anyway.

Please explain in detail, for starters, what kind of system do you have? "right click" on "computer" and select "properties" see if it shows your cpu as the T9300 @ 2.50ghz

If not, what does it say?

Also, your model numbers suffix "ctr" means (configure to order REFURBISHED), so being a refurbished system you can't always be sure whats in them, nor can you know the actual date of manufacture, so be very careful when buying one with this designation.

[mariol90]

When the processor is idle, the clock speed will change. Run a program that will use the CPU (super pi or linx will work) and it should go back to the normal speed.

[rumbero]

I check my clock speed and it was 1190 instead of 2500.
Any ideas how to get it back to at least 2500?

If i remember correctly, the battery is needed as an electric currency buffer to handle high peaks in electric requirements by the CPU. If this can't be supplied, the CPU is usually run with lower currency requirements, which directly translates into lower maximum frequency limits.

So, if you are running this machine on mains and without the battery inserted, than something like this would be the expected behavior. Even if your machine is running connected to the power supply, you should always keep the battery inserted to reach the expected performance maximum of your CPU.

[TuuS]

I assume you're measuring your clock speed with a 3rd party tool? that's an important piece of data you left off, but if so (as mentioned by others) the speed step technology will run the chip slower to save power in some cases.

[Cwrb]

When the processor is idle, the clock speed will change. Run a program that will use the CPU (super pi or linx will work) and it should go back to the normal speed.

Thx for feedback on my question. I now understand the clock speed can be over-clocked, CTR stands for refurbished, and clock speed will be lower when not pushed.

I do not attach battery to save battery life. It lasted 5 years when only attached when using. But it is dead now.

I don't understand how using SuperPI will show CPU clock speed.

I checked my processor with the Intel ID Utility and it indicated the Core 2 Duo CPU T9300 expected speed is 2.5GHz but each core runs at 1.19GHz when not pushed. The 2.5GHz speed is the highest speed each core can operate at.

I would like to know how to test it at its highest speed on both cores and see the result.

[Johan]

I would like to know how to test it at its highest speed on both cores and see the result.

Get RM Clock

Johan

[miro_gt]

so let me give 10 min in trying to explain the whole thing.

every CPU since long time ago supports the Intel SpeedStep technology, which means that the CPU can run on different speed to conserve energy and to keep it at lower temperature. The CPU speed itself is determined by the internal CPU multiplier which produces the CPU operating frequency based on the clock speed that it gets from the clock generating chip in your laptop. As a standard on T61 laptops, that clock speed is 200MHz so the depending on what multiplier the CPU is using at the moment, the CPU speed will vary. For example your T9300 CPU supports multipliers from 6x to 12.5x . Half multipliers are not so common but indeed do exists, so at 12.5 x 200 MHz gives the CPU speed of 2.5GHz, i.e. what your CPU is rated at. The regular multipliers are 6x, 7x, 8x, 9x, etc. to the max supported.

so, there's no need for the CPU to be working at 2.5GHz and use only say 1% of its capabilities when it can work say 5 times slower and use 5% of its capabilities to produce the same result. While it's a bit more complicated, take it that with lower frequency the CPU will use lower energy to operate (thought not by much but still), thus you end up having say better battery life. At the moment when the CPU starts working harder it will raise its internal multiplier to get more speed so that its performance will increase substantially for the task that is doing. The internal multipliers change so fast and so many times that it's hard to actually monitor those when changing, but a program called ThrottleStop is good for showing actual CPU speeds on about 1 sec or so intervals, so that you can get the idea in general. Note the multipliers can change too many times in that 1 sec.

now Intel has different features on different CPUs, but your T9300 supports Intel Dynamic Acceleration (IDA) and Super Low Frequency Mode (SLFM).

What IDA does is it gives you an additional multiplier for one core only for short period of time if you're using programs that is not coded for multi-core use. So indeed from time to time running program like that you'll see your CPU working at 2.7GHz. There's a trick that you can use with ThrottleStop that can lock both cores at IDA, or the so called Dual IDA, so that you lock that extra multiplier on both cores, but I believe that is not subject of this thread. Just to let you know using that trick I run my T9500 at 2.8GHz all the time, instead of 2.6, lol.

What SLFM gives you is that it's able to effectively cut your clock speed on half, i.e. from 200MHz only 100 is supplied to the CPU. This is an effective way to further reduce CPU speed when not needed, as the lowest multiplier of the CPU is usually 6x, so theoretically if there was no SLFM then the lowest speed your CPU can run (or idle actually) will be 6x200MHz = 1.2GHz (same as what you see). However, when using SLFM the CPU can get down to 600MHz. Note that when SLFM is active the CPU actually uses the 8x multiplier as lowest instead of the 6x one (dont ask me why) so you'll actually see down to 800 MHz. Now there's a way to force the 6x multiplier with the SLFM enabled so that you get down to 600MHz, but frankly the difference is not much from 600 to 800 so I'll just leave it like that.

on top of all that, CPUs have different sleep states coded as c3, c6, etc. So if you force the CPU to be running at max speed and doing pretty much nothing, it will put itself in one of those states and still will not use lots of power to run. That's why I said earlier to take the power savings with lower multiplier just like that as you can see now that more things get into the whole picture. The Penryn CPUs are very effective in getting into sleep states fast, thus save more power and dont heat as much as the previous generation of Merom chips (or up to X7900). All CPUs above T7800 (and X7900) are Penryn chips from the T series. X is the extreme CPU in that series chips, i.e. the one with unlocked multipliers.

also, the internal multipliers are locked into all CPUs except the extreme ones, so usually you're not able to use higher multiplier than the maximum one preset in the CPU from Intel, thus can not overclock your chip to run faster. There are ways some people get away with it but that's also different subject.

further, your CPU has heat protection so if it or any other component on the MB (i.e. the northbridge chip) starts getting too hot for some reason then the chipset on the MB will send a signal to your CPU to slow down speed, so it will lock on the lowest multiplier (i.e. 6x) until this signal goes away. This can happen if the heatsink moves a bit from the chip that it sits on, due to movement/vibrations with the laptop. Also the thermal protection of the CPU itself will shut it down completely if it overheats for some reason (i.e. the CPU reaching 100 deg C or so).

anyways, you can see speed bumping up on your CPU when you start actually using it, so run some program and monitor CPU speeds with say CPU-Z (a small utility freeware program).

... looking back I think I wrote most of the whole story, so I'm going to make me a sandwich.



P.S.
- the laptop adapter is sufficient (or is supposed to be) to provide power so that the laptop itself can run with maximum speed with or without battery. It's when on battery that the laptop can slow down to give you longer battery life.

- RMClock is a good undervolting program that has nice CPU monitoring features for you to see how your CPU is working with graphs. Beware however that if you install and activate it, your CPU will work at 2.4GHz maximum as the program itself does not support half multipliers so the maximum multiplier your CPU is going to be using will be 12x (instead of 12.5x). This is because this is old program that lost support some time ago. People switched to ThrottleStop these days.

- if you use the Power Manager, a lenovo software that comes with the laptop, you can have your battery plugged in and not charging eventhough it may not be full. It's a great feature that very few laptops have it. Use it, works awesome. My battery sits at 92% atm and I'm plugged in.

[tyito]

I own several T series laptops. The direct and simple answer is to adjust your power settings in Windows Control Panel. If you have XP, try the max setting. If you have 7, then you could drill down to advanced settings for more granular cpu control.

[TuuS]

Some very good info posted here. Since the subject of battery life has come up, you can set your charge threshold so the battery won't be charged until it reaches a chosen level. I have mine set to charge when it reaches 70% and stop at 95%, and the last time it started to charge was in september, it's currently at 75% and my battery cycle count is at 1. Unless I choose to run it in battery, or am forced to, it will probably be several weeks before it begins to charge again.

You'll need lenovo power manager to set these thresholds, I'm not sure if there are any 3rd party tools that will do this, or if there are any for non-windows applications.

[Cwrb]

Thx Miro_gt, for the great detail.
I'll play with ThrottleStop a while and see what i can accomplish.

[projanen]

Thanks for pulling all of that info together miro_gt. You even mentioned the interesting tangents.

How does this story continue for Core i3/5/7? For instance, Turbo Boost seems very similar to IDA.

[miro_gt]

yes, the turbo boost operates on the same principal as the IDA, with the difference that turbo boost can use more than one multiplier above the rated CPU speed and could do so on each core at the same time at different level.

now it's called turbo boost because your CPU can not work on turbo all the time, because it gets too hot and will most likely overheat. And it gets hot because higher speeds require much bigger voltage increase each time the multiplier goes one step up, compared to the difference between multipliers at the lower end, thus much more current flows through the CPU and that produces more heat. So newer CPUs have thermal limits to which the turbo boost have to comply. In case the thermal limit has been reached, the turbo boost lowers its multiplier to stay within. There is more than one thermal limit, but for simplicity I'll leave it as is. So the cooler your laptop can keep, the higher the turbo boost can go and the longer it can stay there.

( let me try to give an example)
if you have a CPU that say runs on 2.1GHz with 100MHz clock and 21x multiplier, then the same CPU could reach say up to 2.6GHz on turbo with 26x multiplier while going through the rest of the turbo multipliers (22x, 23x, 24x, and 25x as max). Now the higher the speed, the more voltage the CPU needs to distinguish the impulses that go to it, i.e. the information, as those do not come as a perfect rectangular impulse bur rather as a parabolic one, i.e. the shape of a bell, due to the not perfect world that we live in.

This should give you better prospective of the things:
(multiplier) - (CPU voltage required to run it)
..
10x - 0.8000
11x - 0.8125
12x - 0.8250
..
18x - 0.9000
19x - 0.9500
20x - 1.0000
21x - 1.0500 (rated CPU speed of 2.1GHz, i.e. the CPU will not overheat if it remains here, having proper heatsink attached to it)
22x - 1.1000 (turbo boost from here and up, i.e. may or may not sustain during operation)
23x - 1.2000
24x - 1.3500
25x - 1.5000 (2.5GHz)
as you can see from this example (that I just made up, but reflects the reality), at lower multipliers the voltage increase between each step is small, or 0.0125V difference, say between the 10x and the 11x. However at high speed one step up requires much bigger voltage from the previous (ten times bigger), or 0.1500V between the 24x and the 25x. So now you can see why processors usually stay below 3.5GHz or so, as it gets much harder to operate on higher frequency. And this higher voltage requirement is because of the current technology that is used to produce chips, i.e. the silicon transistors and their electrical properties.

Note that the smaller the manufacturing process, the less voltage is needed, but the general picture with the multipliers stay the same. So indeed if you get the CPU in the example above produced at 40nm scale then it will need 1.5000 Volts to operate at 2.5GHz, but if you have the same chip made under 32nm process then it may only need 1.4000 volts to operate at the same level and so will keep cooler temperatures. Thus it may sustain higher turbo boost for longer time. So smaller is better Even it it were to require the same voltage, being physically smaller means the total heat produced will be less.

note that the voltage itself does not equal more heat on the CPU - this is highly dependent on what the CPU actually does. As from my post above, the CPU can put itself into sleep states, thus if you force it to stay at say 2.0 GHz and do nothing then it will stay cooler than if you lock it at say 1.5GHz but utilize 100% of its processing power. I've had a long discussion about which of the two causes more heat, and I believe we agreed to the fact that heat is produces when more circuits inside the CPU are being utilized, i.e. the CPU is doing hard work, rather than when the CPU is not working as hard but using higher voltage to operate.


One of the reasons that I didn't get me a X9000 CPU (unlocked multipliers, so does not need and does not have IDA) for my laptop when I needed more CPU power was due to the fact that the higher multipliers of the X9000 required too high of a voltage jump to operate, as explained above, thus at full CPU utilization the processor would have produced too much heat for my laptops cooling abilities. So I stopped at T9500 at 2.8GHz, which I'm running at 1.0750 V

I've talked to somebody who is using X9000 on his laptop, though not lenovo but same chipset and size (14"), and here's what he said he needs to feed the CPU with:

(speed) - (multiplier) - (voltage)
...
2,6GHz - 13x - 1,0125 V
2,8GHz - 14x - 1,0625 V
3,0GHz - 15x - 1,1375 V
3,2GHz - 16x - 1,2125 V

and he stopped at 3.2GHz because at 3.4 GHz (at about 1.4000 Volts) and full CPU utilization he said the temperatures go above 90 deg C fast and would most likely hit 100 if the CPU keeps working on full, at which point thermal shut down would engage. Now that is too hot.

[nikiiiva]

Hi,
I know it is late answer but someone else could find it useful.
It's been documented and discussed on lenovo's forum before, to get maximum speed you need ether battery to be plugged in, as rumbero sead or to use 90w power adapter without battery. It has something to do with battery acting as capacitor when using low 65w adapter. If you want to know more try searching lenovo forum. You can check realtime cup speed and multiplier with CPU-Z.

All the best