Change your old memory, batterylife benefits! (56k warning)

T4x series specific matters only
Message
Author
jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

Re: Change your old memory, batterylife benefits! (56k warni

#31 Post by jiopsi » Tue Nov 27, 2007 1:18 pm

Nicola wrote:What is 9W?
9 Watts of power, measured with Notebook Hardware Control when in battery mode.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#32 Post by jiopsi » Fri Nov 30, 2007 4:54 pm

Apparently my B1 revision Dothan wasn't as good undervolter as gunston's or aaa's. What revision do you have, you can use CPUz to check it. Here are my tests Banias vs. Dothan.

Banias 1.5GHz 1MB 130nm B1
Image
Dothan 1.6GHz 2MB 90nm B1
Image

1. In idle difference is about 0,25W for Banias.
2. Near idle, the difference gets smaller.
3. In stress it turns to favor Dothan.

- Because Banias is about 10% slower at same speeds, theoretically it consumes more power also near idle, if the task given is something like superpi which can be processed faster.
- Clearly my Banias was great undervolter sample and Dothan not so great sample, although it gets 0,7V in 6x.
Last edited by jiopsi on Fri Nov 30, 2007 5:36 pm, edited 1 time in total.

aaa
ThinkPadder
ThinkPadder
Posts: 1062
Joined: Fri Jun 08, 2007 2:36 pm

#33 Post by aaa » Fri Nov 30, 2007 5:29 pm

How strange.

I am using a C0 (which needs hacks to work), pretty sure gunston is using a B1.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#34 Post by jiopsi » Fri Nov 30, 2007 5:33 pm

NHC stress test passed about same as gunston's, but Superpi did need more voltage. What hacks does C0 need?

aaa
ThinkPadder
ThinkPadder
Posts: 1062
Joined: Fri Jun 08, 2007 2:36 pm

#35 Post by aaa » Fri Nov 30, 2007 5:35 pm

jiopsi wrote:NHC stress test passed about same as gunston's, but Superpi did need more voltage. What hacks does C0 need?
It basically needs NHC I think (or is it RMClock?), ie it won't work properly unless you manually set the voltages.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#36 Post by jiopsi » Fri Nov 30, 2007 5:38 pm

Manually every time you start? Or just once?

aaa
ThinkPadder
ThinkPadder
Posts: 1062
Joined: Fri Jun 08, 2007 2:36 pm

#37 Post by aaa » Fri Nov 30, 2007 5:39 pm

Well apps like NHC start automatically, so just once.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#38 Post by jiopsi » Fri Nov 30, 2007 5:51 pm

Ok, so no problems there. Maybe I will try change newer revision, if I can find it couple of dollars. Do you know if LV Dothans like 758 are supported in BIOS?

aaa
ThinkPadder
ThinkPadder
Posts: 1062
Joined: Fri Jun 08, 2007 2:36 pm

#39 Post by aaa » Fri Nov 30, 2007 6:04 pm

jiopsi wrote:Ok, so no problems there. Maybe I will try change newer revision, if I can find it couple of dollars. Do you know if LV Dothans like 758 are supported in BIOS?
I don't think it'll matter; the reason the C0 doesn't work is because it's not supported in BIOS, but it is as you can see still usable.

I also think it won't necessarily be "better", as only the soldered ULV models really push the limits. In fact, the 758 is probably soldered as well. If you can find one on the cheap, go ahead, but don't go out of your way for it.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#40 Post by jiopsi » Fri Nov 30, 2007 6:10 pm

Do you get somekind of error messages or how do you know it's not supported in BIOS? Just found un-soldered 758 C0 about 10€, so if I can win that on ebay I will be testing that.

All in all, I think normal, LV and ULV models are all the same, just cherrypicked and named in factory.

EDIT: found that 855PM chipset doesn't support LV or ULV according to Intel http://compare.intel.com/pcc/showchart. ... ture=en-US

aaa
ThinkPadder
ThinkPadder
Posts: 1062
Joined: Fri Jun 08, 2007 2:36 pm

#41 Post by aaa » Fri Nov 30, 2007 6:34 pm

jiopsi wrote:Do you get somekind of error messages or how do you know it's not supported in BIOS?
It boots up stuck at 600mhz.
EDIT: found that 855PM chipset doesn't support LV or ULV according to Intel http://compare.intel.com/pcc/showchart. ... ture=en-US
That sounds silly. LV was around before the 915.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#42 Post by jiopsi » Fri Nov 30, 2007 6:40 pm

aaa wrote:
EDIT: found that 855PM chipset doesn't support LV or ULV according to Intel http://compare.intel.com/pcc/showchart. ... ture=en-US
That sounds silly. LV was around before the 915.
Yeah, I though that also, so maybe I will give the 758 a try.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#43 Post by jiopsi » Tue Dec 04, 2007 5:09 pm

Found some newer 256MB memory and tested it, memory usage was 209MB with both, and I don't know if it matters in how many % it is from whole capacity. I will be testing even newer memory which should consume about same with 512MB if the specs are comparable.
Image

With 512MB Mosel -02
Image
With 256MB Hynix -04
Image

- 0.4W in idle is about right
- It grows to ~0.95W in SuperPI
- About same in Ram stress test
- So, specs are quite comparable, and seems that only IID0 and Standby matters
- Again, don't know if % of used memory matters, but I will be testing with even newer 512MB memory (with about same specs), so we will see if it matters.

Nicola
Sophomore Member
Posts: 136
Joined: Wed Nov 21, 2007 6:54 pm
Location: Italy

#44 Post by Nicola » Wed Dec 05, 2007 12:46 pm

I ask excuse you. I do not speak English and I have read this fastly post; I have decided the same to write this, I hope is "useful".

Banias
normal 1,39 V
minimal frequency 0,96 V

Dothan
normal 1,340 V
minimal frequency 0,988 V

If the cpu always work to the minimal speed then Banias consumes little.
If they always work to the maximum speed then Dothan consumes little.
If it is set up normal, depends on the work load. Obviously if the higher voltage works much consumes little Dothan using. If it works little it consumes little Banias.

aaa
ThinkPadder
ThinkPadder
Posts: 1062
Joined: Fri Jun 08, 2007 2:36 pm

#45 Post by aaa » Thu Dec 06, 2007 10:52 am

Some ram tests. I'm not getting very good results :? .

Code: Select all

8.9w "Old" Hynix 256MB (8 chip)
8.9w "Old" Elpida 512MB (16 chip)
9.0w "Old" Kingston/Promos 512MB (16 chip)
9.1w "Old" 2x512 (elpida + kingston)
9.0w "New" Crucial/Micron-335 512MB (8 chip)
9.1w "New" Crucial/Micron-335 2x512
9.1w "New" Crucial/Micron-335 1x1gb (16 chip)
9.1w "Old" Kingston/Promos 1x1gb (16 chip)
So apparently it isn't as simple as "old" vs "new". Perhaps the Mosel stuff was particularly awful? In any case I'll try to get that same Hynix 512 and see.

And it appears the amount of ram (sticks or MB) doesn't matter much for some reason. I presume it's because most of it on standby or something.

jiopsi
Sophomore Member
Posts: 145
Joined: Tue Nov 20, 2007 5:02 pm
Location: Salo, Finland

#46 Post by jiopsi » Thu Dec 06, 2007 5:02 pm

Yeah, it definitely isn't easy as how old it is, because manufacturing process may have been same like years. Mosel was 140nm and never are definitely smaller, it would be nice if manufacturers would reveal this information. The specs seems quite combarable though, so it would be good to make decisions based from that.

Did you use NHC to check those wattages, if so remember to use about same battery % level in tests, as voltage drops when battery gets empty. Did you test in stress?

Post Reply
  • Similar Topics
    Replies
    Views
    Last post

Return to “ThinkPad T4x Series”

Who is online

Users browsing this forum: No registered users and 5 guests