Skip to main content

Unsolved

This post is more than 5 years old

Closed

sfmarra

26 Posts

183459

December 8th, 2005 19:00

B130 memory issue

It appears that the b130 bios is only reporting my memory speed at 400mhz and it should be 533mhz.  I've verfied the memory stick is a 533mhz.  Has anyone else experience this?  Is the a fix for this?
 
Thanks 

sfmarra

26 Posts

December 13th, 2005 20:00

thanks

chadteck

20 Posts

December 13th, 2005 20:00

Ahains,
 
I can tell you where the northbridge (chip that contains the memory controller) is located.  My friend and I took his notebook apart together since he doesn't care, and I don't know if I'm keeping mine yet.  Once you have removed the keyboard, you can see the packaging of the northbridge (but not the chip which has the actual model number on it).  The best way to explain it that is is under where the keys "QWAS" are on the keyboard.  You will see the green chip packaging partially sticking out from underneath the laptop case plastic.  You must remove the casing to actually see the labelling on the chip. FW82910GML, which corresponds to the 910GML chipset, is exactly what was on his chip. 
 
I was wondering the same thing about why they would put 533MHz FSB processors in a 400MHz motherboard.  The only thing I can think of is that they have a surplus of Pentium M 740's or that the 740's are just cheaper than the 735's now.  I would have rather gotten a 400MHz FSB Pentium M as one of the things that bothers me about this situation is that the CPU FSB and RAM are running asynchronously, which typically results in lower performance than even a synchronous FSB at the lower speed (400MHz FSB, 400MHz RAM speed)...  It irritates me that Dell mentions the 915GM chipset in the manual for the B130, but apparently doesn't use it.  I am guessing they had originally planned to use it, but decided to use the 910GML to cut costs and hoped consumers wouldn't notice.

sfmarra

26 Posts

December 13th, 2005 20:00

chad,
do you have a link to the manual that states this should be the 915 in the b130??  If dell isn't providing something they are adversiting,  with the hopes people wont notice, that might get them into a little bit of trouble.  I don't think the AG office would like that to much! ;)

erniecars

26 Posts

December 14th, 2005 04:00

Here’s my take.

First off, I don’t know how much info cpu-z can accurately provide in this case due to the proprietary hardware and possibly even a customized chipset. Maybe the next version could more accurately reflect the info, and maybe it will work with better with whatever is preventing us from seeing memory info on several dell models, not just the b130. I have a b130 and a 6000 sitting here with the same exact issues: Memory running at 400mhz while 533 was advertised by Dell, and memory info in cpu-z is scrambled.

I also read a rumor that dell is retagging b120s as b130s.  There is no way to tell by the chipset model number exactly what fsb speeds or memory bus speeds are supported because intel is tweaking their stuff all the time. One example was the 810 chipset, some of which supported 66, 100, 133, or any combination of those depending on what revision it was. Very confusing and dangerous considering older revision boards would fry cpus that worked fine in newer revision boards with newer chipsets.

The inspiron 6000 has 2x 512 samsung ddr2 533 running at 400. Unless I’m reading the specs wrong on this model, this should mean the “effective” fsb is 800mhz, which basically solves the problem right? (I mean the potential speed issue but not Dell’s false advertising)

So I put the two dimms from the 6000 into the b130 (which previously had 1x 256mb hynix ddr2 533 running at 400) and it ran dual channel as well. So this appears to be a solution that I can accept unless I am way off base here about a dual channel configuration.

Message Edited by erniecars on 12-14-2005 12:32 AM

pg60

17 Posts

December 14th, 2005 12:00

erniecars i think you are right that we need to figure out what the actual performance hit is.

i am not the hardware geek i used to be and don't know much about the effect of asynchronous FSB: memory speed ratio. Back when I used to overclock my desktop cranking up memory speed > FSB speed was not good.

In this case FSB speed > memory speed so I guess it depends on the efficiency of async mode in the 910GML chipset. Since 533 DDR2 is kind of new it looks looks like lots of late 2004/early 2005 laptops were configured with 533 FSB & 400 memory. So maybe it is not that bad.

I am trying to be optimistic because this is xmas present for my son and I am running out of time here. If I return it there is no comparable laptop i can find at same pricepoint.

My B130 shoould return from it's road-trip to Dell today (they told me to return it for "repair", then apparently did nothing but say my 533 meomory runs just fine at 400).

I am going to use Sandra and do lots of benchmarking to check out performance before I decide what to do. Unfortunately I bought single 1GB PC2-4200 sodimm from newegg so the best I can do for dual mode is my new 1GB + the 256MB that came with it. Does dual mode require same size sodimms? I will probably benchmark it with 1GB & with 1.25 GB.

Message Edited by pg60 on 12-14-2005 08:05 AM

erniecars

26 Posts

December 14th, 2005 13:00

Yeah you have to have same size dimms, and preferably the exact same brand and model as well. I was also wondering, does everyon else with a 1.73 have their chip running at 133 fsb before multiplication? A cheaper 1.6 chip runs at 200 on the inspiron 6000 (speed step is not causing this)

ahains

28 Posts

December 14th, 2005 13:00

erniecars,
Yes, my Pentium M 740 shows a FSB of 133mhz.
Wouldn't 200mhz FSB be impossible with a 533mhz bus speed, since you would need a fractional multiplier?
I'm not sure that this 133mhz number really matters though, doesn't this just refer to the crystal (or whatever they use instead of a crystal)? I mean I don't think anything actually communicates at 133mhz, right?
The PCI bus is still 66mhz (IIRC), memory <=> memory controller is 400mhz, cpu to chipset is 533mhz, and core speed is 1.73ghz (speedstep notwithstanding).
 
The memory has a 2:3 FSB:DRAM ratio, so it is running at 200mhz (400mhz since I guess it does two operations per cycle). I think if you had a 200mhz crystal the memory would just be running at 1:1 FSB:DRAM ratio, giving you the same memory speed.
 
Or maybe I'm totally misunderstanding :)
-Adrian

ahains

28 Posts

December 14th, 2005 13:00

pg60,
I haven't seen any 100% conclusive rules on what you need to run in dual channel mode, but I think you essentially need two sticks that are the same size and that can handle whatever speed you're running. Not sure what happens if they have different CAS..
 
This page on dell's site mentions they must be "same size, speed, etc", but I'm not sure what the etc refers to :)
 
I was on the fence when I ordered my new memory from newegg a few days ago. I finally decided that I was not likely to ever go higher than 1gb of ram with this laptop, so I ordered (2) 512mb sticks so I can run in dual channel mode. It was essentially the same price, so I figured might as well.
I was recently looking at various benchmarks, you should really only see about a 5% perf increase by running in dual channel mode in any real world usage. The synthetic memory benchmarks will show twice the throughput, but that is only useful for when you want to run synthetic memory benchmarks ;)
 
-Adrian

erniecars

26 Posts

December 14th, 2005 15:00

I guess 133 is normal for a 533fsb, as 100 is normal for 400fsb, 200 for 800fsb etc. We are basically dealing with a setup similar to old p4 desktop setups, where the cpu ran 400 or 533 fsb and the memory was pc133 sdram or, ddr at 266, 333, or 400. ddr being just double the base clock of 133, 166, or 200. This makes me wonder if you could really squeeze much more performance out of the b130 by having the memory at 533. I kindof doubt there will me a noticable difference between single and dual channel given the same amount of memory at 400mhz. As it is, the base clock speed of the fsb is 133 and the memory speed is 200 so what would 533 ram have a base clock speed? I think the ram is capable of 266 (double data rate being 533) so perhaps this is still just beyond the capability of the chipset and again we have been duped. Sorry Im kinda rambling.
 
Anyway, just the fact that the manual states that this model may use the 915 chipset, i still cannot rule out the possibility that they ran out of 915 mainboards so they made a new bios for the b120 that makes it look like a b130, since apparently the 910 does support speedstep and 533fsb cpus. That rummor just might be true afterall. If the b130 never had a 915, why did the manual say it may? and if it didnt what model did have the 915? Im stumped here.

Message Edited by erniecars on 12-14-2005 11:05 AM

ahains

28 Posts

December 14th, 2005 15:00

erniecars,
The memory running at 400mhz with the 133fsb has a divider of 2:3 (133/2 ~= 66; 66*3 ~= 200mhz).
For the memory to run at 533mhz, it would just use a divider of 2:5 (66*4 ~= 266mhz);
 
533mhz is a 33% increase over 400mhz, so I think the performance difference would be significant. Of course since my B130 doesn't support 533mhz RAM I can't compare compilation times or benchmarks :(
 
Note that if you look at real life memory benchmarks, increasing the clock speed of the memory has a significantly larger performance boost than going to dual channel mode at the original speed.
 
-Adrian

chadteck

20 Posts

December 14th, 2005 16:00

Some good points have been brought up, and more questions raised. Regarding the negative impact that this motherboard/cpu setup has on performance, I believe that the theoretical performance hit is significant. But, as I have stated before, the real-wrold performance hit, may be insignificant. I have run a few benchmarks and have found that the memory bandwidth for my laptop is just over 2500MB/s. Ram with a clockspeed of 400MHz is theoretically capable of 3200MB/s. Due to not being able to acheive a 100% efficieny of the memory controller, you wont ever see the theoretical as the actual (but you can see close to it in an Athlon 64 *cough*). From experience, a typical setup running the memory syncrhonously at 400MHz should yield a memory bandwidth of say, 2800-3000MB/s. Again, this number depends on how efficient the memory controller is. Having an asynchronous FSB:DRAM frequency hurts the efficieny of the memory controller, and lower overall bandwidth. If you compare the figure I have gotten from benchmarks with an esimated actual performance if the system were able to run the memory at 533MHz, it seems to be quite significant. The theoretical bandwith of DDR2 at 533MHz is ~4250MB/s. Even assuming a poor efficieny, this should equate to a minimum of 3500MB/s actual performance. This means that we are most likely being deprived of 1000MB/s plus of memory bandwidth, and possibly more since I have assumed low for the efficiency when running the memory at 533MHz. I have seen benchmarks of a 400MHz FSB Pentium M and 400MHz memory where the system had just near 3000MB/s of memory bandwith. This suggests that there is approximately a 500MB/s performance hit in this case just for running the CPU and RAM asynchronously. Couple that with the performance gain that would be had by running the memory at 533MHz, and the difference is quite significant.

As far as how this effects real-world performance though, I would say that overall, you probably won't "notice" it. The importance of memory bandwidth depends on the task being performed, and it is hardly ever important outside of intensive tasks like gaming (this is the wrong laptop for that anyways), file compression, encoding, video /large photo editing. And, as I have said before, due to the architecture of the Pentium M, it seems that memory bandwidth is not very important as long as it's decent. The Pentium M's relation to the Pentium 3 seem to make it less memory dependent (the P3 was designed when the average memory bandwidth was way lower than what we are seeing today); also a large on chip cache (which the Pentium M has) helps to reduce the negative impact that sub-par memory bandwidth has on system performance.

I hope that my answer some questions, but I still encourage everyone to keep arguing with Dell. I still do not agree with what they have done and how their techs are dismissing the issue.

Edit: To clarify gaming comment. This laptop is fine for non-3D gaming, and will probably run some newer 3D games decently, but for good 3D performance, a laptop with a dedicated video card that has hardware 3D acceleration features would be much better.

Message Edited by chadteck on 12-14-2005 12:57 PM

pg60

17 Posts

December 14th, 2005 16:00

thanks to both of you guys, i am trying to follow along :-)

it has become obvious that i am never going to get the memory running at 533Mhz.

as much as i am frustated by dell, i am trying to take a step back and review my goals for this laptop. it is x-mas present for my son, not a gaming or development rig, wanted to get him strong foundation with Pentium M 740 CPU & 1GB of memory and I wanted a 15.4 widescreen for DVD watching.

I paid 460 + shipping + tax + 59 for the 1GB of RAM. So I am into this for < 600 which is a pretty good deal. If i return it I don't think I can find this good of a deal anywhere else by x-mas.

so the question in my mind is how much will this problem affect my son's use of this computer which will be:

internet
internet games (not much 3D)
watching dvds
school work

and i am figuring probably not that much. i just wish I had stayed with the celeron and saved $100 since I am stuck with celeron mobo. In fact I might try that angle with Dell, refund me $100 for the Pentium upgrade and swap it for a celeron 360.

sfmarra

26 Posts

December 14th, 2005 16:00

I think for what want to do with it, you'll be more than fine.  Also with 1 gig or ram, im sure he'll be able to play alot of 3d games.   I think alot of us are just upset because we were told to expect one thing, and dell did a switch-a-roo on us.  Probably hoping no one would really take notice. Its not that im not satisfied with the computer, because I am. I'm pretty much persuing this on principal at this point.  I dont believe they can legally mislead therr consumers and I intend to get what i paid for.

pg60

17 Posts

December 14th, 2005 18:00

sfmarra,

i had to shut mine down before my son gets home from school.

not sure what you want to know but i ran Sandra Lite and

- memory SPD information is detected, it is PC2-4200 DDR2 533MHz, i got timings for 266MHz & 200MHz

- FSB is running at 533MHz (4 * 133)

- memory is runn at 400MHz (2 * 200)

- chipset is Dell (not Intel!) 915GM/915GMS/910GML

i hope that helps some.

sfmarra

26 Posts

December 14th, 2005 18:00

http://www.lavalys.com/products/overview.php?pid=3&lang=en

forgot the link for everest.  its a free trial

 

thanks,

Was this post helpful?

View All

No Events found!

Top