Does Intel's Turbo Boost activate during booting?

Does it have any effect on boot time?

 

nope...ur SSD or HDD has an effect.. nver heard of turbo boost doing that.. its more for apps.

 

I hope it wouldn't activate during boot, as it doesn't know the usage conditions yet...

 

I would assume it does: the usage conditions are that the cores are cool enough to allow X amount of turbo boost (depending how many cores were being used) and give that amount of boost until either the temp's increased over a predefined threshold and/or boost wasn't needed by the O/S anymore.

 

Turbo Boost is a hardware solution built inside the CPU. It doesn't require any drivers to run, so it is active during booting etc. I don't think it has any impact on boot time though. If I am wrong, please correct me .

 

I think it should have a huge impact on boot time - have you tried booting with a PII300 lately?

 

No. If the cpu is bottlenecking boot times, why does going from a HDD to an SSD have such a tremendous impact?

 

Both can impact boot times - they are not mutually exclusive.

 

Agreed. Both have an impact, while the HDD/SSD has the bigger impact (assuming one has a recent CPU).

 

This statement has some sense, and is a reasonable, albeit very general claim.

However, this statement is wrong on so many levels. Turbo Boost simply will not have an appreciable effect on boot time. I cannot stress this enough: if the CPU is not the bottleneck, a greater clock will have no discernible impact on that metric of performance. You can think that it should, have faith that it will, or pray all you want, but that's not going to change anything.

As for the Pentium II 300 reference, that is a baseless comparison that serves only to demonstrate that the CPU *can* influence such a metric. This in no way justifies the previous statement and is actually a logical fallacy, which one might attribute to the hasty generalization. While the statement is perhaps true for one case, that does not mean it applies to all. Specifically, while one CPU may present a true bottleneck, a Pentium II 300 is hardly representative of all CPUs, especially those created within the last decade (where the CPU is highly unlikely to represent the bottleneck during boot for any conventional operating system).

Therefore, such a statement is misguiding and illogical, and can only serve to propagate misinformation. I am not attacking you; I just value accuracy and rational thought.

 

Consider yourself corrected.

 

thats wierd turbo boost shud b inbuilt with the chip. thats y u can en/disable in bios?

 

I agree. Thing is, the CPU is a bottleneck at some moments during booting.

I didn't believe this at first but I had some guys proving it to me with a desktop that could be hardware locked to lower multipliers. The lower multiplier had a an effect on boot time. (I can dig up the link but it is in dutch).

 

That's a driver for Intel's integrated GPU, with incredibly confusing naming - it isn't at all related to the CPU's Turbo Boost. Turbo Boost needs no driver - as the person you quoted said, it runs at CPU level completely outside of the OS.

 

Hmm interesting. I stand corrected then.

 

There's no driver like this in my Windows (7 Pro 64-bit) device manager, however I do have a CPU with Turbo Boost technology (i7-720QM). The feature is working without any issues.

 

The difference it would make to your boot time would be so miniscule that it would probably not even be measurable. Given that turboboost doesn't make that much difference in gaming it would make even less difference to boot time, for the most part of booting up, the processor is doing virtually nothing, it doesn't really have to do much work seeing as all that's really happening is that the operating system is being loaded from the hard drive into the RAM.

 

No it would not use Turbo Boost, you wouldn't want it to.
Remember when all cores are in use, Turbo boost cannot take place. When booting, all cores are utilized to boot up windows thus not allowing for any margin for Turbo Boost to enable without causing the TDP to go higher then specified.

Also Turbo Boost is a OS function when:

Intel, here.

Edit:
This is also why some laptops/desktops with load controlled fans will speed up dramatically during start up (you get an loud) fan noise, that's because for often then not the processor is under near full load on all cores.

 

When the O/S is initializing itself, the faster CPU will finish faster, period.

Booting up Windows is not simply copying what is on the HD to the Memory.

I'm not sure how many cores Windows can use to boot - but when it is able to utilize a hardware supported feature like Turbo Boost (it definitely does not need an O/S running to initiate it) because some functions are only using a single core, then it will boot faster.

Some things that may make it look like it doesn't make a difference are drivers that wait a specific amount of time to initiate an no matter how fast the CPU is: then, that time period will always be the same.

And the quote by Crimsoned does not take into account that the boot process by default requests the highest processor performance state 'P0'.

 

I agree the name could have been better but it does have something to do with turbo boost. The driver basically balances the TDP's of the CPU and GPU via Intels Intelligent Power Sharing.

Yes, it will work without any driver but the above quote seems to suggest it might work better in the above case. Anyone interested enough to test it?

 

Well i am well aware the cores are set to highest state but that mean none of the cores are in c0 which means turbo boost is not possible with out windows controlling the cpu states. Remember turbo boost only works if atleast one of the cores is in c0.
Also the 100 mhz increase in clock would account for maybe tenths of a second.
Read intels pdf on the architecture and features.

 

The cores are not set to the highest state - the chip is. What state each individual core is set to is the job of the scheduler.

And yes, tenths of a second is what makes things faster...

 

In a consumer-grade Intel system, the cpu is always constrained first of all by the i/o capabilities of the chipset and the hard drive and then the bandwidth of the memory system (as driven by the chipset or an on-cpu die mmu). Always.

So attempting to run the cpu 'faster' during boot doesn't buy you much (if anything at all). You need to streamline/minimize the amount of software and drivers that load during startup to have even a remote chance of improvement.

Problem is, you're likely to need all of those drivers and much of that software shortly after boot up anyway.

So what to so, what to do.........

Other than to fully understand the boot process of a wintel system and stop wasting time arguing about it in a web board thread, perhaps doing nothing is the best course of action.

 

C0 is when the core is active, perhaps you mean C3/C6. Core speed at POST BIOS will depend on the BIOS.

When all cores on my i7 860 are active turbo is still possible, just that it will be limited to one turbo bin. With 2 or more cores inactive (C3/C6) that means a higher turbo bin can be used as long as turbo is enabled and selected and TDP or temperature is not exceeded.

 

You are definitely wrong there (the cpu is not always constrained by the i/o of the chipset and HD).

Sometimes, the things the cpu needs to do can always be done faster on a higher clocked cpu and the i/o of the chipset and/or HD do not come into play.

 

:sigh: In what systems and circumstances will the cpu not be waiting for i/o during the wintel boot process and can a human being actually see it without instrumentation?

 

Ive done test before on my amd pIIi quad between core # and frequency to figure out the typicalminimum for the average user. I found the booting times to be similar at the appropriate core #
4 cores 3.6ghz, 2ghz, 1.6ghz
2 cores at 2.0ghz, 3.6ghz
1 core 1.6 ghz, 2.0ghz, 3.6ghz

 

Crimsoned, can you expand on this and make it a little more clear what it is you found? Thanks.

 

This is not entirely true. When there is enough thermal headroom, Turbo Boost can increase the clock of all cores by 1 bin (133MHz) while all cores (or threads) are loaded in 100%.
I tested it and it does work that way. I loaded my CPU with BOINC. The application loaded all 8 threads. The CPU was running at 1,73GHz (while standard clock is 1,6GHz).

 

getting off track guys. The original question was about TB during boot and what could reasonably be done to speed up that process.

Not how TB behaves when a machine is already running and the TB drivers have been loaded by the OS.

 

newposter, you seem to be conveniently forgetting that TB is a hardware cpu feature - it does not rely on an O/S to activate it.

 

noooo, TB is both hardware in the cpu and an OS-specific driver. The OS can take charge of TB through the defined Intel API and supplied driver code.

 

Here's a run with turbo on/off, interestingly a single core with HT outperforms 4 cores with HT. My guess would be that it's a while before Windows starts using c-states so only one bin for 4 cores (3.5GHz) but 5 bins for enabling just one core (4.1GHz). Other than that, nothing exciting. 5 runs with times in seconds.

Code:

Cores	Threads	Turbo	Run 1	Run 2	Run 3	Run 4	Run 5
4	8	No	24.94	24.27	24.45	25.14	24.66
4	8	Yes	24.28	23.92	24.09	24.3	23.95
1	2	Yes	23.02	22.14	22.53	22.23	23.09

Non Turbo = 3360MHz
Turbo = 3520MHz – 4160MHz

Here's what I used to measure the boot time, it's not ideal but very easy to use and small enough to just copy to the startup folder.

 

Thanks for that Dufus.

Yes, today's processors are more than fast enough (for booting).

Yes, the actual time saved is rather miniscule (again, for booting).

But also, Yes, Turbo definitely is active during bootup.

I really can't understand why some thought otherwise?

The CPU is what does the 'work', after all... one that is doing it faster will accomplish more in the same time frame, or, the same amount of work in a smaller time frame.

 

wow sweet booting times...
Operating system/storage device ? (SDD im guessing?)

 

ok ppl , why are we arguing about turbo boost? Nothing great about it and i certainly doubt it helps as much as an SSD in boot time.. thats the punchline ..

 

To sum up the OP's questions.

1. Does Intel's Turbo Boost activate during booting?

It can do but is dependent on how the BIOS hands over to the OS.


2. Does it have any effect on boot time?

Very small effect. Using multiple turbo bins with multiple cores requires use of C3/C6. As there is a lot of time spent waiting on I/O the CPU might regularly sleep in C6 which has a high exit latency, so what is gained by the higher turbo frequency can be lost by the higher wake up latency. It may even be possible for the initial boot to the desktop to be slower with turbo and C3/C6 than without turbo and no C3/C6. The difference is tiny though and I would think most laptops will probably have C3/C6 enabled by default anyway regardless if turbo is enabled or not.

No SSD in the above test, just a single HDD. The OS however is just used for testing so has next to nothing installed on it which is just as well as I didn't want to spend too much time rebooting.