AMD64-based laptops from HP/Compaq are crippled

[alex goldman]

I've been researching my next laptop purchase, and whether it will run
Linux, and completely accidentally stumbled upon an alarming issue that
people need to be more aware about.

Apparently (and if my understanding is correct), HP/Compaq disables
dual-channel memory access in Socket 939 motherboards, making CPU <-> RAM
communication slower, and overall performance less than what people would
expect when buying a system with the outward characteristics of what they
think they are getting.

http://zv6000forums.com/viewtopic.php?t=148

 

[NoNoBadDog!]

I've been researching my next laptop purchase, and whether it will run
Linux, and completely accidentally stumbled upon an alarming issue that
people need to be more aware about.

Apparently (and if my understanding is correct), HP/Compaq disables
dual-channel memory access in Socket 939 motherboards, making CPU <-> RAM
communication slower, and overall performance less than what people would
expect when buying a system with the outward characteristics of what they
think they are getting.

http://zv6000forums.com/viewtopic.php?t=148

Since they are shipping with DDR and not DDr2, what would be the purpose of
enabling the dual channel?

Have you been to the AMD website and reviewed their white papers for the
AMD64 platform, and why they do *NOT* recommend DDR2 RAM?

There is no perceivable increase in performance with DDR2 over DDR in a 64
bit platform.
The margin of difference is so small as to be insignificant.

You should probably review the benefits of the AMD64 platform
(hypertransport, on board memory controller, etc), that give the platform
it's speed.

Bobby

 

[Ed Light]

Since they are shipping with DDR and not DDr2, what would be the purpose
of enabling the dual channel?

My dual channel AMD gets a bit less than twice the memory speed in memtest86
when running dual channel.

No AMD runs DDR2.

Why they would disable dual channel I don't know. Since they can completely
configure the pc with compatible memory, it doesn't make sense, unless they
randomly throw whatever cheap memory they can find into it. Then it figures.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.

 

[Suburban]

Since they are shipping with DDR and not DDr2, what would be the purpose
of enabling the dual channel?

Dual channel works with plain DDR

 

[alex goldman]

Since they are shipping with DDR and not DDr2, what would be the purpose
of enabling the dual channel?

It would appear you are asking a question here, showing your confusion of
"dual channel" and DDR2, but then (below) you tell me to read AMD "white
papers", as if you have. Why would you do that?

 

[Robert Klute]

I've been researching my next laptop purchase, and whether it will run
Linux, and completely accidentally stumbled upon an alarming issue that
people need to be more aware about.

Apparently (and if my understanding is correct), HP/Compaq disables
dual-channel memory access in Socket 939 motherboards, making CPU <-> RAM
communication slower, and overall performance less than what people would
expect when buying a system with the outward characteristics of what they
think they are getting.

http://zv6000forums.com/viewtopic.php?t=148

The laptops use the Turion chip. According to AMD, the Turion's
integrated memory controller can use PC1600, PC2100, PC2700 or PC3200
unbuffered DDR SO-DIMMs and has a maximum throughput of 3.2 GB/s.

So, explain to me what engineering in dual channel support is going to
buy you. BTW, does any vendor manufacture dual channel compatible
SO-DIMMs?

 

[alex goldman]

The laptops use the Turion chip.

Wrong. _Some_ of the laptops use the Turion chip (quite few as of this
writing, actually).

Regardless ...

According to AMD, the Turion's
integrated memory controller can use PC1600, PC2100, PC2700 or PC3200
unbuffered DDR SO-DIMMs and has a maximum throughput of 3.2 GB/s.

Do you understand how dual-channel access works? You have 3.2 GB/s per
memory module, i.e. up to 6.4 GB/s. Dual-channel access is more, not less,
important when slow memory is used.

So, explain to me what engineering in dual channel support is going to
buy you. BTW, does any vendor manufacture dual channel compatible
SO-DIMMs?

There is something you have in common with the NoNoBadDog

 

[Robert Klute]

Wrong. _Some_ of the laptops use the Turion chip (quite few as of this
writing, actually).

The zv6000, the laptop under discussion here and in the forum you
referenced, uses the Turion chip.

Regardless ...


Do you understand how dual-channel access works? You have 3.2 GB/s per
memory module, i.e. up to 6.4 GB/s. Dual-channel access is more, not less,
important when slow memory is used.

NO, the Turion's memory controller is 64-bit and limited to a max of
3.2GB/s. Dual channel won't help. Read the specs. Here is the
executive's graphic for you:
http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_12651_12663,00.html

 

[alex goldman]

The zv6000, the laptop under discussion here and in the forum you
referenced, uses the Turion chip.

Wrong.

Go to hp.com,

http://www.shopping.hp.com/webapp/s...0_series&catLevel=2&tab_switch=true&tab=specs

you'll get

Choices:
AMD Sempron(TM) 3000+ (1.8GHz/128KB L2 Cache)
AMD Sempron(TM) 3200+ (1.8GHz/256KB L2 Cache)
AMD Athlon(TM) 64 3200+ (2.0GHz/512KB L2 Cache)
AMD Athlon(TM) 64 3500+ (2.2GHz/512KB L2 Cache)
AMD Athlon(TM) 64 3800+ (2.4GHz/512KB L2 Cache)
AMD Athlon(TM) 64 4000+ (2.4GHz/1MB l2 Cache)


May I ask why you insist on spreading misinformation *after* you've been
corrected?

NO, the Turion's memory controller is 64-bit and limited to a max of
3.2GB/s. Dual channel won't help.

Wrong. Turion is a Socket 754 CPU, so it doesn't even support dual-channel
memory access. The discussion is about the CPUs that do.

 

[Robert Klute]

The zv6000, the laptop under discussion here and in the forum you
referenced, uses the Turion chip.


NO, the Turion's memory controller is 64-bit and limited to a max of
3.2GB/s. Dual channel won't help. Read the specs. Here is the
executive's graphic for you:
http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_12651_12663,00.html

And just to be absolutely clear, only the chips with 128-bit memory
controllers can use dual channel memory and ALL the mobile AMD-64 chips
are listed as being socket 754 and having 64bit memory controllers.


http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_10220_10221^11030,00.html


As for the forum thread you referenced, the posters are just a bunch of
whiny idiots. Since only an idiot would think that a laptop would use
an 89W 939 chip in it. Talk about a hot system, it would have to come
with a nomex pad to keep you from getting 2nd degree burns.

 

[Robert Klute]

Wrong.

Go to hp.com,

http://www.shopping.hp.com/webapp/s...0_series&catLevel=2&tab_switch=true&tab=specs

you'll get

Choices:
AMD Sempron(TM) 3000+ (1.8GHz/128KB L2 Cache)
AMD Sempron(TM) 3200+ (1.8GHz/256KB L2 Cache)
AMD Athlon(TM) 64 3200+ (2.0GHz/512KB L2 Cache)
AMD Athlon(TM) 64 3500+ (2.2GHz/512KB L2 Cache)
AMD Athlon(TM) 64 3800+ (2.4GHz/512KB L2 Cache)
AMD Athlon(TM) 64 4000+ (2.4GHz/1MB l2 Cache)


May I ask why you insist on spreading misinformation *after* you've been
corrected?

OK, so I am the idiot, HP is using the 939 chips. Damn, how big is the
auxillary fan unit you have to lug around with it? If you plan on using
the CPU, disk and display at the same time, you better get the 120W
brick instead of the 90W version.

 

[J. Clarke]

Wrong.

Go to hp.com,

http://www.shopping.hp.com/webapp/s...0_series&catLevel=2&tab_switch=true&tab=specs

you'll get

Choices:
AMD Sempron(TM) 3000+ (1.8GHz/128KB L2 Cache)
AMD Sempron(TM) 3200+ (1.8GHz/256KB L2 Cache)
AMD Athlon(TM) 64 3200+ (2.0GHz/512KB L2 Cache)
AMD Athlon(TM) 64 3500+ (2.2GHz/512KB L2 Cache)
AMD Athlon(TM) 64 3800+ (2.4GHz/512KB L2 Cache)
AMD Athlon(TM) 64 4000+ (2.4GHz/1MB l2 Cache)


May I ask why you insist on spreading misinformation *after* you've been
corrected?

Now, if you go to the AMD site you'll find that the Mobile Athlon 64 is
_also_ listed as having 3.2 GB/sec memory support.

Now, it may be that HP is not using Mobile chips but instead using the
desktop chips, in which case I presume that they also provide a SNAP
reactor to run the thing.

Do not assume that all Athlon 64s are created equal.

Further, since the memory controller on the Athlon 64 is build into the
chip, one has to work pretty hard to disable dual-channel support.

And as another poster pointed out, are there any 128-bit SODIMMS out there?
If not then even if dual-channel support _was_ present you would need to
have two SODIMMS in order to use it.

Wrong. Turion is a Socket 754 CPU, so it doesn't even support dual-channel
memory access. The discussion is about the CPUs that do.

So please show us the manufacturer's specifications for an available (i.e.
you can buy it somewhere) laptop that indicate that it has a 939 pin
socket.

 

[General Schvantzkoph]

I've been researching my next laptop purchase, and whether it will run
Linux, and completely accidentally stumbled upon an alarming issue that
people need to be more aware about.

Apparently (and if my understanding is correct), HP/Compaq disables
dual-channel memory access in Socket 939 motherboards, making CPU <-> RAM
communication slower, and overall performance less than what people would
expect when buying a system with the outward characteristics of what they
think they are getting.

http://zv6000forums.com/viewtopic.php?t=148

Turion's are in 754 packages not 939, they have one memory controller not
two. If you want performance then make sure you get a processor with a 1M
Cache, the cache size is much more important then either clock speed or
dual memory channels. I have a Compaq R3000z laptop with an Athlon 64
3400+ (1M cache, socket 754) and a desktop with an Athlon 64 3800+ (1/2M
cache, socket 939). The 3400+ is twice as fast as the 3800+ on some tasks.

 

[alex goldman]

Now, if you go to the AMD site you'll find that the Mobile Athlon 64 is
also listed as having 3.2 GB/sec memory support.

Has it not been made abundantly clear to you what chips are used in the
laptops in question?

Further, since the memory controller on the Athlon 64 is build into the
chip, one has to work pretty hard to disable dual-channel support.

It's disabled in BIOS. HP admits as much. The motherboard is socket 939. Why
don't you read the site and maybe think for a moment.

 

[alex goldman]

Turion's are in 754 packages not 939,

What do Turions have to do with it?

they have one memory controller not
two. If you want performance then make sure you get a processor with a 1M
Cache, the cache size is much more important then either clock speed or
dual memory channels.

For some tasks. For others, dual-channel will run twice faster.

 

[General Schvantzkoph]

What do Turions have to do with it?

You're right it's a 939, someone else mentioned the Turion which would
make sense for a notebook so I assumed that's what this one had. Where did
you get the idea that it only uses one of it's memory controllers? I don't
see any mention of that on HP's website, did I miss something?. As long as
you get two identical DIMMs it should run dual channel.

For some tasks. For others, dual-channel will run twice faster.

I can't think of any task that you would run on a notebook where the dual
channels will make a difference. I can give you an example of the type of
task where cache size matters,

http://www.polybus.com/linux_hardware/index.htm

 

[Wes Newell]

For some tasks. For others, dual-channel will run twice faster.

That's pure BS, expect for bandwidth benchmarks. Overall performance
increase for dual over single channel is 5% at most, with some apps
actually running slower, and some running faster. Although I'll admit that
building a 939 MB without support for both channels is just flat stupid.

 

[Wes Newell]

Has it not been made abundantly clear to you what chips are used in the
laptops in question?


It's disabled in BIOS. HP admits as much. The motherboard is socket 939. Why
don't you read the site and maybe think for a moment.

I've come into this late, but are you saying they sell a laptop with
socket 939, and only provided for one memory controller, leaving the other
lines for the second controller from the cpu socket not connected? Only a
complete moron would do this. The dual channel ram is basically the only
advantage over a socket 754 cpu. And it doesn't provide much advantage
performance wise until you get to dual core. My guess is that this isn't a
939 board. I couldn't find anything to indicate it is. Just because the
bios says second channel disabled doesn't mean a thing. they can put
anything they want in the bios. Now if in fact they did use 939 just so
you could run the later cpu's, but have the second memory controller not
connected I can say one thing. I wouldn't buy it.

 

[alex goldman]

Overall performance
increase for dual over single channel is 5% at most

Where did you get this idea?

 

[alex goldman]

I've come into this late,

Easy come easy go.