12.5" HP Elitebook 2570P Owner's Lounge

[Tech Inferno Fan]

INFO: 2570P public profile is increasing

Our HP 2570P is now #25 (on the uprise - highest it's been) and Lenovo X230 #22 (on the decline) on N**'s top 50 most popular laptops. In Oct 2012, when they banned me for recommending a 2570P, the Lenovo X230 was #13 and 2570P had neither a review on that site nor any entry in that chart. See for yourself:

Best Laptops - The Top 50 Most Popular Notebook Computers (now) and in Oct-2012

Had the 2570P been fairly represented it would have always been above a Lenovo X230 in stature because it's a signficantly better piece of hardware. Now HP has their premium Haswell ZBooks with 'workstation performance' features which we got with our 2570P as a standard feature. Wonder if X230 owners with buyer's regret at being given biased buying info by N** could take legal action?

This news is filtering out further into the public. Ebay-US prices of 2570P units are on the rise.

[bjorm]

I used the CPU reported TDP rather than temps. Thats because temps will vary depending on the ambient temperature, any other heatsink/fan/enclosure mods the user has and whether they have additional devices active generating internal heat (HDDs, WWAN, Bluetooth). Too many variables prevent doing a meaningful temperature comparison.

FYI: right now my ambient temperature where I am in Australia is 30-35 degrees. Yes, it's hot. In Europe you'd be what hitting 0 degrees around this time??

Yes, at this time here is something around 0 degrees

Im really dissapointed because I was making myself trouble with 45W TDP, this last one 3632QM was really good and cool. Now, even with i5 I have average temps around 45C even with gelid gc extreme - its disgrace for me

[Tech Inferno Fan]

Yes, at this time here is something around 0 degrees

Im really dissapointed because I was making myself trouble with 45W TDP, this last one 3632QM was really good and cool. Now, even with i5 I have average temps around 45C even with gelid gc extreme - its disgrace for me

Can you please do a table with multi and TDP and post it like I've done? I'll be able to see if you really just have one of the very inefficient CPUs. It might serve to help others perhaps avoid the 45W i7-36xxQM CPUs.

As you know, our system was designed with a cooling system for 35W CPUs. We have some margin to play and can improve cooling but IVB is a die shrink so turbo boost of inefficient CPUs can see a huge temp spike. I countered that spike by added an extra copper shim between the heatsink and CPU. A decent, but not crazy expensive thermal paste helps to quickly move heat away from the CPU to the copper.

You've decreased the fan speed to quieten down the system so have decreased the cooling efficiency. No surprise then that a efficient 35W i7-quad is your preferred CPU.

EDIT: I just played with Window's Power management settings. If I set maximum CPU performance to 99% it disables turbo and only allows up to x27 non-turbo multiplier. So you'd need Throttlestop if you wanted to decrease the turbo of a 45W CPU by say two to four multipliers, bringing TDP down to 35W i7-quad levels.

[bjorm]

other question: why all the time throttlestop shows temps on CPU something around 8-10C less than hwinfo?

at the same moment we can see:

[Tech Inferno Fan]

other question: why all the time throttlestop shows temps on CPU something around 8-10C less than hwinfo?

at the same moment we can see:

[ATTACH]10039[/ATTACH]

Good question. Something to raise with unclewebb, Throttlestop's creator: http://forum.techinferno.com/throttlestop-realtemp-discussion/23-throttlestop-guide.html#post39

[unclewebb]

Exit ThrottleStop, delete the ThrottleStop.INI configuration file and then run ThrottleStop again. This will force it to read the correct TJ Max value from your CPU. You should always do this after swapping CPUs to make sure that it uses the correct value. If there is still a difference then open up the configuration file for HWINFO and make sure it is using the same value as ThrottleStop is using. When TJ Max is not correct, the reported core temperatures will not be correct. When both programs are using the same TJ Max values, the reported temperatures should be close to identical.

The 3rd Gen CPUs are generally set to 105C.

[bjorm]

Correct, I've already changed tjmax in throttlestop from 98C to 105C and differences are only about 2C.

btw, I got info from my 3632QM buyer, he sent me a information about VID and TDP from my previous CPU, and what a mess! I sold so good CPU

x12

VID: 0,80V

TDP: 10,1W

Do You think there are possible differences for VID and TDP if test is conducting with 65W adapter? I suppose until doing max stress test there shouldn't be differences for x12, am I right?

[Tech Inferno Fan]

Correct, I've already changed tjmax in throttlestop from 98C to 105C and differences are only about 2C.

btw, I got info from my 3632QM buyer, he sent me a information about VID and TDP from my previous CPU, and what a mess! I sold so good CPU

x12

VID: 0,80V

TDP: 10,1W

Do You think there are possible differences for VID and TDP if test is conducting with 65W adapter? I suppose until doing max stress test there shouldn't be differences for x12, am I right?

35W CPU has no issues running with 65W AC adapter. Even 45W CPU has no issue using a 65W AC adapter other than there's a HP popup complaint and it runs a bit hot if pushing the system under constant load.

Your candidate i7-3632QM has a x12 voltage of 0.80V and TDP=10.1W. It's pretty much the same as my 'efficient' i7-3740QM running x12 at 0.8105V. Meaning if you could source the same high grade i7-3740QM I had then you could multiplier-limit it with Throttlestop to run like a cool i7-3632QM or unleash it to get an extra 1200pts in 3dmark11 physics score (7000->8200). If there's no much price difference between the two in your local market then certainly would be worth scouting for. Those i7-37xxQM CPUs also can gain an extra 400Mhz on gaming systems so they are probably going to have higher demand, and price, for them. Then again, i7-37xxQM CPUs are more likely to appear in the market as gamers upgrade to XM CPUs.

I've also updated my results for full 4-decimal place voltage readings here. Did you check your i7-3630QM to see what voltage it runs at x12? If it's significantly higher then it would explain the higher temps you saw all round. It would be helpful if you could publish that and the same table as I created for your i7-3630QM. That way we can have a public knowledgebase that would help 2570P owners wanting the create the coolest running, highest performance system.

I'll be adding the x12 voltage alongside the CPU listings of owners on the first page if they volunteer that info.

[bjorm]

I've also updated my results for full 4-decimal place voltage readings here. Did you check your i7-3630QM to see what voltage it runs at x12? If it's significantly higher then it would explain the higher temps you saw all round. It would be helpful if you could publish that and the same table as I created for your i7-3630QM. That way we can have a public knowledgebase that would help 2570P owners wanting the create the coolest running, highest performance system.

I'm going to do this comparison table with 3632QM and 3630QM when I get my new 3632QM. I hope it's gonna be this week.

Thank for Your advice with CPU's but really I can't see any improvement with better CPU's than 3632QM. I believe than even 37** or 38** won't give me better utilization of GPU in eGPU than 3632QM. Take a look for CPU load, CPU Package Power (the same values as TDP in throttlestop) and GPU utilization.

Look how good GPU is loaded with the same level of CPU utilization and how big differences we have in TDP at the same time

red: i5-3230M

green: i7-3630QM

red: i7-3632QM

[Tech Inferno Fan]

I'm going to do this comparison table with 3632QM and 3630QM when I get my new 3632QM. I hope it's gonna be this week.

Thank for Your advice with CPU's but really I can't see any improvement with better CPU's than 3632QM. I believe than even 37** or 38** won't give me better utilization of GPU in eGPU than 3632QM. Take a look for CPU load, CPU Package Power (the same values as TDP in throttlestop) and GPU utilization.

Look how good GPU is loaded with the same level of CPU utilization and how big differences we have in TDP at the same time

red: i5-3230M

green: i7-3630QM

red: i7-3632QM

[ATTACH]10045[/ATTACH]

Great. I'll be completing the table with the efficient i7-3740QM which I hope too arrives this week. Adding your info as well will give us a good spread of what i7-quad CPUs to get and which to not if want the coolest running system.

To help that along, I've added an x12 VID entry alongside the 2570P owner's CPU details on the first page. Right now we know about (mine and yours). Any other 2570P owners in that table want to volunteer theres? Details of how to do that are here.

I assume that Green = i7-3630QM. Yes, for your plots there is little difference in eGPU performance of the 35W i7-3632QM and 45W i7-3630QM. What we do however see is the 45W one runs an average 7.49W higher TDP during that testing. That's your extra heat right there.

As you probably know, jot123 will be getting an AMD card HD78xx or better card comparising aginst his GTX660. He'll be producing similar awesome at-a-glance plots. I believe the result will show that some games do better with AMD and others with NVidia. Having *both* cards giving ability to choose the right weapon of choice for your game of interest.

[bjorm]

As you probably know, jot123 will be getting an AMD card HD78xx or better card comparising aginst his GTX660. He'll be producing similar awesome at-a-glance plots. I believe the result will show that some games do better with AMD and others with NVidia. Having *both* cards giving ability to choose the right weapon of choice for your game of interest.

Yes, we are going to do some more gameplay comparisons. At the time, I've been waiting for my GTX680 warranty respond (not working fan) and I've been browsing internet for 7970 to compare them in Battlefield4 and Crysis3. And I don't know why this old GPU is so expensive - far more than notreferent GTX680

After tests I will probably sell GTX680 if doesn't ensure better gameplay than GTX670.

[Tech Inferno Fan]

Yes, we are going to do some more gameplay comparisons. At the time, I've been waiting for my GTX680 warranty respond (not working fan) and I've been browsing internet for 7970 to compare them in Battlefield4 and Crysis3. And I don't know why this old GPU is so expensive - far more than notreferent GTX680

After tests I will probably sell GTX680 if doesn't ensure better gameplay than GTX670.

I've read that the bitcoin miners are heavily invested in HD7950 and HD7970, both giving great mining performance per watt of power consumed, creating a high market demand for them.

It might be worthwhile looking at the newer R9 AMD GPUs instead as better value options for gaming: http://forum.techinferno.com/diy-e-gpu-projects/5550-most-optimal-gpu.html#post77508

[jacobsson]

INFO: How to choose the most efficient 45W i7-quad

Here my contribution for the 3820QM:

EDIT: To be honest the TDP pretty much impossible to determine since it jumps up and down like crazy. The Voltage is stable though,

[Tech Inferno Fan]

Here my contribution for the 3820QM:

[ATTACH]10058[/ATTACH]

More good stuff with your i7-3820QM ES a reasonably efficient chip. Though can you check your x25 and x29 result? The voltage is higher than the previous multiplier for both yet you report the TDP being lower. This doesn't make sense since the higher frequencey and voltage sees the the CPU doing more work within the a time unit. Each should therefore have a TDP higher than the previous multiplier.

Good to see yours holds x34, same as my efficient i7-3740QM. Must be your 8MB cache (vs 6MB on mine) that explains your 3dmark11 8.5k physics score where I got 8.2k.

[jacobsson]

More good stuff with your i7-3820QM ES a reasonably efficient chip. Though can you check your x25 and x29 result? The voltage is higher than the previous multiplier for both yet you report the TDP being lower. This doesn't make sense since the higher frequencey and voltage sees the the CPU doing more work within the a time unit. Each should therefore have a TDP higher than the previous multiplier.

Good to see yours holds x34, same as my efficient i7-3740QM. Must be your 8MB cache (vs 6MB on mine) that explains your 3dmark11 8.5k physics score where I got 8.2k.

Regarding the TDP: It jumps up and down like crazy so I don't think there be any meaning to it, meaning no accuracy? I'm not sure if you guys have stable TDP when measuring?

[Tech Inferno Fan]

Regarding the TDP: It jumps up and down like crazy so I don't think there be any meaning to it, meaning no accuracy? I'm not sure if you guys have stable TDP when measuring?

The way I get a pretty stable TDP is I start a x12 TS 1024M 4-core/8-thread benchmark, let it settle into a rhythm and not do not stop it. Then I just change the multi along to the next one .. x23, wait a few secs for it to stabilize and take the volt & TDP reading, then change multi to x24, wait a few secs for it to stabilize and take the volt & TDP reading and so on until the I gather the results for all available multipliers.

I found this gave accurate numbers that can be repeated with subsequent runs hours or days apart.

[jacobsson]

The way I get a pretty stable TDP is I start a x12 TS 1024M 4-core/8-thread benchmark, let it settle into a rhythm and not do not stop it. Then I just change the multi along to the next one .. x23, wait a few secs for it to stabilize and take the volt & TDP reading, then change multi to x24, wait a few secs for it to stabilize and take the volt & TDP reading and so on until the I gather the results for all available multipliers.

I found this gave accurate numbers that can be repeated with subsequent runs hours or days apart.

That sounds reasonable! I'll post a new table later, although the voltage should be correct.

[Tech Inferno Fan]

2570P surpasses X230 at N** top 50 list

As pointed out a few days ago, our systems seem to be getting a lot more interest. In fact we see it is now #23 (highest ever) and Lenovo X230 is now #26 (lowest ever?) on N** top 50 list. We now see the market correctly positioning these systems relative to each other's feature set. 2570P is higher likely because the public now recognizes it's superior build and upgradeability in a compact package. See for yourself: Best Laptops - The Top 50 Most Popular Notebook Computers

Manufacturers should take note. 'thin-and-light ultrabooks' are way too limited for power users. They are a consumable throwaway item. Unfortunately in Haswell, no manufacturer has provided a system comparable to our 2570P (12.5", expresscard slot for eGPU, upgradeable CPU, dual SSD/HDD or SSD/HDD+ODD).

Not that it matters.. we see a i7-3820QM upgraded 2570P giving similar performance to a i7-4700MQ Haswell system. The 29% battery life savings touted as a Haswell feature are mostly confined to the ultrabooks using the UHT platform (CPU+chipset+iGPU on a chip). The power Haswell systems (eg: 15" Macbook Pro) with separate chipset and CPU see only 14% battery life improvements over Ivy Bridge systems like our 2570P. Those numbers based on Apple's official battery estimates for IVB and Haswell comparable systems with identical Wh battery capacity.

[bjorm]

Yes, This is correct.

		x12	x13	x14	x15	x16	x17	x18	x19	x20	x21	x22	x23	x24	x25	x26	x27	x28	x29
i7-3630	VID	0,85	0,85	0,86	0,87	0,88	0,89	0,9	0,92	0,93	0,95	0,96	0,97	0,98	1	1,01	1,03	NA	NA
	TDP	10,6	11,2	11,9	12,8	13,7	14,4	15,4	16,4	17,4	18,4	19,6	20,7	22	23,4	24,9	26,4
i7-3632	VID	0,83	0,83	0,83	0,83	0,84	0,85	0,86	0,87	0,88	0,9	0,92	0,93	0,95	0,97	0,99	1,01	1,03	1,05
	TDP	10,2	10,8	11,4	11,9	12,7	13,4	14,4	15,3	16,4	17,3	18,6	19,8	21	22,3	24,1	25,7	27,6	29,8

[ATTACH]10068[/ATTACH]

[ATTACH]10069[/ATTACH]

[Tech Inferno Fan]

Yes, This is correct.

@bjorm , any reason why you didn't take the i7-3630QM up to the x32 4-core turbo frequency? Intel Core i7 3630QM Notebook Processor - NotebookCheck.net Tech . If it only turbos up to 2.7Ghz then it's behaving like a i7-3612QM Intel Core i7 3612QM Notebook Processor - NotebookCheck.net Tech . In any case, I've updated the table with your data at http://forum.techinferno.com/hp-business-class-notebooks/2537-12-5-hp-elitebook-2570p-owners-lounge-37.html#post77432. You may note I'm using full 4 decimal places for the voltage.

I'm thinking the x27 voltage/TDP is now a better reflection of the CPU efficiency. We see the premium i7-37xxQM and i7-38xxQM are more efficient chips than the 35W i7-quads when evaluating the higher frequencies. Though some of these have higher initial x12 voltage/TDP.

[bjorm]

I need to check 3630QM one more time. Cause after swap from 3630QM to 3632QM I wasn't able to run benches (x12 to top) If I didn't restart throttlestop. Maybe the same happened when I got x27 on 3630QM (on the picture You can see during top 3630QM setting, throttle sees only 27x).

[Tech Inferno Fan]

I need to check 3630QM one more time. Cause after swap from 3630QM to 3632QM I wasn't able to run benches (x12 to top) If I didn't restart throttlestop. Maybe the same happened when I got x27 on 3630QM (on the picture You can see during top 3630QM setting, throttle sees only 27x).

Ensure EIST is selected, something I noticed in your last screenshots is disabled, and your "Turn On" Throttlestop. If you have a genuine i7-3630QM then it has up to 3.2Ghz 4-core turbo boost per Intel Core i7 3630QM Notebook Processor - NotebookCheck.net Tech . Though we are finding that on our 2570P that will be decreased by 100Mhz (x31) or 200Mhz (x30) depending on how efficient your particular CPU is.

[bjorm]

I put today 3630QM back but there is something wrong with throttlestop. When EIST is selected, I run for few test and there was higher TDP comparing to yesterday and the multiplier was one higher than estimated. After restart TDP was the same like yesterday, with EIST selected. So I'm afraid this data from throttlestop is not stable to make any comparisons

[Tech Inferno Fan]

I put today 3630QM back but there is something wrong with throttlestop. When EIST is selected, I run for few test and there was higher TDP comparing to yesterday and the multiplier was one higher than estimated. After restart TDP was the same like yesterday, with EIST selected. So I'm afraid this data from throttlestop is not stable to make any comparisons

If you delete the throttlestop.ini (in Throttlestop dir) and just start it in monitoring mode, do a 1024M test, then it should then show the highest turbo multiplier and it's TDP. That will provide one more piece of important data.

[bjorm]

	Tmax (TS b / 3dm11)	3dm11 PS		x12	x13	x14	x15	x16	x17	x18	x19	x20	x21	x22	x23	x24	x25	x26	x27	x28	x29	x30	x31	x32
i7-3630QM	91 / 83	6906	VID	0,85	0,85	0,86	0,87	0,88	0,89	0,9	0,92	0,93	0,95	0,96	0,97	0,98	1	1,01	1,03	1,06	1,07	1,09	1,11	1,12
	max rpm - 5100		TDP	10,6	11,2	11,9	12,8	13,7	14,4	15,4	16,4	17,4	18,4	19,6	20,7	22	23,4	24,9	26,4	28	29,9	32	33,9	36,8
i7-3632QM	85 / 74	6784	VID	0,83	0,83	0,83	0,83	0,84	0,85	0,86	0,87	0,88	0,89	0,91	0,93	0,95	0,96	0,98	1,01	1,03	1,05	NA	NA	NA
	max rpm - 4500		TDP	10,2	10,8	11,4	11,9	12,7	13,4	14,4	15,3	16,2	17,3	18,4	19,8	21	22,3	24,1	25,7	27,6	29,8	NA	NA	NA

For the same settings (ambient temp, AC adapter, etc), 1024M benchmark took 320 sec with 3632QM and 308 sec with 3630QM, 4% difference. Even less difference in PhysicsScore 3dm11. In my opinion there is no point of getting better CPU's. :)