Disappointing upgrade: Sempron 2200+ --> Phenom II 4X 840

[Rodney Pont]

I was looking for something with graphs, but you're right,
memtest86+ will give the answer. I was thinking maybe the
shape of the graph would be important.

Being colour blind I prefer the AIDA 64 benchmarks:

http://www.aida64.com/product/aida64-extreme-edition/overview

The blue section at the left shows the performance, the faster it is
the longer the blue section.

 

[Darklight]

you should have looked at cpu passmark

cpu passmark
for $109.95 £85.52 JBO solutions phenom x4 840 3.2 ghz rank 190
for $109.99 £96.99 Ebuyer phenom x4 960T 3.0 ghz rank 160
for $109.99 £83.99 overclockers uk amd FX-4100 x4 3.6 ghz rank 141

Phenoms AM3
FX AM3+

second question whats your PSU

 

[Darklight]

for £60 to £65 Athlon x4 640 AM3 rank 193 cpu Passmark

 

[Krypsis]

Wanting more raw computing power, I went from Sempron 2200+
in ASUS Asus A7V400-MX (socket A) to Phenom II 4X 840 in Asus
M4N68T-M V2 (socket AM3).

I used to have a Sempron 2800+. It was a dog! Just about anything would
run rings around it. I gave the whole computer away a year or two back
after having it in storage. It is still in use today. Did put a braver
video card in it for the new owner though.

I must say I am quite disappointed. I was hoping for about 10X speed
boost even with applications that can only use a single core. After all,
more than six years separate the two (expected boost suggested
by Moore's law would be 64X). The real life tests show nothing
of this sort! Not even close to 10X.

I upgraded to a @6600 Quad in a Gigabyte G33M-DS2R motherboard with an
NVidia 8500GS video card (now Nvidia GeForce GT430) and 2 Gig of RAM
(now 8 Gig). It was way overkill for what I do these days.

For my day to day stuff, I use a Pentium 4 HT @ 3 GHz with 4 GB of RAM
and that seems to cope more than adequately with my workload. The
Quadcore rarely even gets turned on

DIVX encoding using VirtualDub and DivX 5 codec became
faster by only about 3X. OK, this may also have something to do
with the video card change (does it? - I don't really know; went
from dedicated nVIDIA Quadro FX 3000, 256 MB to integrated
nVIDIA GeForce 7025, 512 MB - not sure what is supposed to be
better; I need occasional hardware stereo so Quadro will be
going back into the slot).

To make sure it's not GPU, I also tested using a program that does
no graphics (multiparametric fitting anf FFT; essentially raw numbers
crunching). Alas, a fairly common task that took 23 min on the
6.5 years old cheap system now took 9 min - only ~2.5X faster.

On board graphics are always a bit of a dog though there have been some
significant improvements in recent years. I will always favour external
graphics adaptors even if, in my usage, it only provides for snappy
screen updates.

Granted, programs that can use SMP run correspondingly faster
(about 3.3X over just using a single core) but still... Are my results
observations fairly typical? If not, what gives? This $200 upgrade
pales in comparison with the $160 upgrade 6.5 years ago when I
went from dual Celerons 366 in ABit BP6 to the Sempron 2200+.
There, the speed boost was of near cosmic proportions.

Any comments? Thanks!

- Dima

I have found that what I need to be done fast is handled adequately by
my P4. For the rest, I have plenty of time so I can afford to wait. I
could overclock my Quadcore and get an even greater speed boost. In
fact, that was one of the reasons I picked this particular CPU as it has
loads of headroom but, apart from some experiments when new, it has
always run at a standard clock speed.

If I was into some serious gaming, I daresay the Quadcore would be more
useful.

 

[DK]

Mine has the 6MB L3, and the 7Zip score is 8.6MB/sec. using 4/4 CPU threads. I
watch all 4 cores in Task Manager and
they are like synchronized swimmers at the Olympics.
I love the decompression rate of 135MB/sec.

What version of 7Zip do you have? The old 4.6 beta only
uses 33% of the 4 cores and they are definitely not sybchronized.

 

[DK]

you should have looked at cpu passmark

cpu passmark
for $109.95 £85.52 JBO solutions phenom x4 840 3.2 ghz rank 190
for $109.99 £96.99 Ebuyer phenom x4 960T 3.0 ghz rank 160
for $109.99 £83.99 overclockers uk amd FX-4100 x4 3.6 ghz rank 141

Phenoms AM3
FX AM3+

I thought that AM3 offers more choices of inexpansive mobos and CPUs.
Also, I got the 840 for $86. I like keeping things frugal.

second question whats your PSU

Nexus Value 430. I bought it for its inaudible fan. Has been rock solid in
the previous setup and all voltages in the current are as expected and
stable.

Dima

 

[SC Tom]

What version of 7Zip do you have? The old 4.6 beta only
uses 33% of the 4 cores and they are definitely not sybchronized.

4.65. It doesn't say Beta, so I assume it's the standard released version. Which reminded me to check for an update.
I'll let you know how 9.20 goes once I install and test with it.

 

[DK]

4.65. It doesn't say Beta, so I assume it's the standard released version.
Which reminded me to check for an update.
I'll let you know how 9.20 goes once I install and test with it.

Even more weird. 4.65 is what I had before and it was definitely not
using all cores. I now tried 9.2 and in it I can at least see options that
Paul said he had in 4.60. In the 9.2m it has "Number of CPU threads"
but bor some reason only 1 and 2 shows in the pulldown list. The CPU
utiization with 2 threads is 40% and consequently the compression
rate is a little faster but nowhere near yours. And we have identical
MB and RAM, same OS (XP SP3, right?) and very similar albeit
not identical CPU.

 

[SC Tom]

What version of 7Zip do you have? The old 4.6 beta only
uses 33% of the 4 cores and they are definitely not sybchronized.

I installed 9.20 and ran the benchmark on it. Pretty much the same as before- 8.7MB/s and all 4 cores synchronized
within 3 or 4% of each other. It would start at ~60%, then ramp up to 100% and stay there until the next pass. Memory
used was 851MB (forgot to mention that on the last reply).

I ran the 1GB "create a 7z file" as in one of the other sub-threads. Mine took 3 min. 35 sec. to complete, with the 4
cores fairly balanced from 48-53%. Towards the end of the compression, core number 1 (I think) dropped off to about 10%.
But that may have been because OE6 was checking for new mail. The other cores picked up the slack, increasing usage by
about 7-8%.

I did mistakenly start creating a .zip file instead of a .7z one and it took considerably longer, even though it still
used all 4 cores, fairly balanced across them all. I quit it after 6 or 7 minutes with it being about 50% done. Don't
know why there would be such a disparity in time, unless 7ZIP doesn't create WinZip files that well. Is that possibly
what you did that took 18.5 minutes?

 

[SC Tom]

Even more weird. 4.65 is what I had before and it was definitely not
using all cores. I now tried 9.2 and in it I can at least see options that
Paul said he had in 4.60. In the 9.2m it has "Number of CPU threads"
but bor some reason only 1 and 2 shows in the pulldown list. The CPU
utiization with 2 threads is 40% and consequently the compression
rate is a little faster but nowhere near yours. And we have identical
MB and RAM, same OS (XP SP3, right?) and very similar albeit
not identical CPU.

Yep, I'm running XP Home SP3. The only "tweaking" I've done to my system is unlocking the other two cores (like I posted
earlier), and changing my RAM timing from 9-9-9-25-34 to 8-8-8-21-31, but I don't see where that would make as much
difference as there seems to be between our two systems. The only other real differences I see are that I'm running an
add-in video card (GT 240) and SATA HDD instead of PATA. Do you have the latest BIOS and other drivers from the ASUS
site?

Here's a greatly snipped text report from CPU-Z for comparison, if you want:

CPU-Z version 1.58

Processors
-------------------------------------------------------------------------
Number of processors 1
Number of threads 4

Processors Information
-------------------------------------------------------------------------

Processor 1 ID = 0
Number of cores 4 (max 4)
Number of threads 4 (max 4)
Name AMD Phenom II X4
Codename Deneb
Specification AMD Phenom(tm) II X4 B55 Processor
Package Socket AM3 (938)
CPUID F.4.3
Extended CPUID 10.4
Brand ID 29
Core Stepping RB-C3
Technology 45 nm
TDP Limit 160 Watts
Core Speed 3215.1 MHz
Multiplier x FSB 16.0 x 200.9 MHz
HT Link speed 1004.7 MHz
Instructions sets MMX (+), 3DNow! (+), SSE, SSE2, SSE3, SSE4A, x86-64, AMD-V
L1 Data cache 4 x 64 KBytes, 2-way set associative, 64-byte line size
L1 Instruction cache 4 x 64 KBytes, 2-way set associative, 64-byte line size
L2 cache 4 x 512 KBytes, 16-way set associative, 64-byte line size
L3 cache 6 MBytes, 48-way set associative, 64-byte line size

Chipset
-------------------------------------------------------------------------
Northbridge NVIDIA GeForce 7025 rev. A3
Southbridge NVIDIA nForce 630a rev. A2
Graphic Interface PCI-Express
PCI-E Link Width x16
PCI-E Max Link Width x16
Memory Type DDR3
Memory Size 4096 MBytes
Channels Dual, (Unganged)
Memory Frequency 669.8 MHz (3:10)
CAS# latency (CL) 8.0
RAS# to CAS# delay (tRCD) 8
RAS# Precharge (tRP) 8
Cycle Time (tRAS) 21
Bank Cycle Time (tRC) 31
Command Rate (CR) 1T
Uncore Frequency 2009.4 MHz

Memory SPD
-------------------------------------------------------------------------
DIMM # 1
SMBus address 0x50
Memory type DDR3
Module format UDIMM
Manufacturer (ID) G.Skill (7F7F7F7FCD000000)
Size 2048 MBytes
Max bandwidth PC3-10700 (667 MHz)
Part number F3-10666CL9-2GBXL
Number of banks 8
Nominal Voltage 1.50 Volts
EPP no
XMP no
JEDEC timings table CL-tRCD-tRP-tRAS-tRC @ frequency
JEDEC #1 6.0-6-6-17-23 @ 457 MHz
JEDEC #2 7.0-7-7-20-27 @ 533 MHz
JEDEC #3 8.0-8-8-22-31 @ 609 MHz
JEDEC #4 9.0-9-9-25-34 @ 685 MHz

DIMM # 2
SMBus address 0x51
Memory type DDR3
Module format UDIMM
Manufacturer (ID) G.Skill (7F7F7F7FCD000000)
Size 2048 MBytes
Max bandwidth PC3-10700 (667 MHz)
Part number F3-10666CL9-2GBXL
Number of banks 8
Nominal Voltage 1.50 Volts
EPP no
XMP no
JEDEC timings table CL-tRCD-tRP-tRAS-tRC @ frequency
JEDEC #1 6.0-6-6-17-23 @ 457 MHz
JEDEC #2 7.0-7-7-20-27 @ 533 MHz
JEDEC #3 8.0-8-8-22-31 @ 609 MHz
JEDEC #4 9.0-9-9-25-34 @ 685 MHz


Display Adapters
-------------------------------------------------------------------------
Display adapter 0
Name NVIDIA GeForce GT 240
Revision A2
Codename GT215
Technology 40 nm
Memory size 512 MB
PCI device bus 2 (0x2), device 0 (0x0), function 0 (0x0)
Vendor ID 0x10DE (0x1043)
Model ID 0x0CA3 (0x8328)
Performance Level Default
Core clock 135.0 MHz
Shader clock 270.0 MHz
Memory clock 135.0 MHz
Performance Level 2D Desktop
Core clock 405.0 MHz
Shader clock 810.0 MHz
Memory clock 324.0 MHz
Performance Level 3D Applications
Core clock 550.0 MHz
Shader clock 1340.0 MHz
Memory clock 1700.0 MHz

 

[Paul]

Even more weird. 4.65 is what I had before and it was definitely not
using all cores. I now tried 9.2 and in it I can at least see options that
Paul said he had in 4.60. In the 9.2m it has "Number of CPU threads"
but bor some reason only 1 and 2 shows in the pulldown list. The CPU
utiization with 2 threads is 40% and consequently the compression
rate is a little faster but nowhere near yours. And we have identical
MB and RAM, same OS (XP SP3, right?) and very similar albeit
not identical CPU.

Just out of curiosity, have you looked at your BIOS screen recently ?

First, you should have "full screen logo" disabled, in case the
motherboard presents an image instead of text. (A couple of my
motherboards default to presenting the full screen logo, so this
has to be disabled.)

Next, I'd want to check the BIOS declaration of the processor identity.
Is the processor mis-identified, or is the model information and
frequency right ?

Either 7ZIP is only offering "1" and "2" as options, because the program
can only handle two threads of execution. (Some algorithms can't be
"divide and conquer" indefinitely.) Or, the program might be offering
those options, because it thinks the processor only has two cores.
And it might get that information from the operating system.
I understand as well, from watching Linux boot screens (dmesg), that
the BIOS passes information about the number of cores in some kind
of table. So it might be possible for the BIOS to mis-inform the OS.
This wouldn't be a problem, if the OS also had its own identification
procedures. So it's a matter of whether the OS places all its trust in BIOS
tables, or whether it also does some of its own detection.

For example, when I boot Linux in a virtual machine on my PC, Linux
complains that the BIOS table "reports one core" which is correct,
but "the core number is 1 instead of 0", implying the virtual BIOS
isn't passing "core0" as the identity of the virtual processor. That's
how I know that at least with Linux, Linux is inspecting some info
from a BIOS table, and in that case, did not like what it saw. Linux
didn't crash or anything, and the message was more of a warning than
an error. It didn't actually affect the operation of the OS.

Paul

 

[DK]

Just out of curiosity, have you looked at your BIOS screen recently ?

Sure. More than I initially planned to, in fact

First, you should have "full screen logo" disabled, in case the
motherboard presents an image instead of text.

Hate this and disable right away.

Next, I'd want to check the BIOS declaration of the processor identity.
Is the processor mis-identified, or is the model information and
frequency right ?

All seems to be perfectly correct.

Either 7ZIP is only offering "1" and "2" as options, because the program
can only handle two threads of execution. (Some algorithms can't be
"divide and conquer" indefinitely.) Or, the program might be offering
those options, because it thinks the processor only has two cores.

Specificaly, this is how it looks:

Number of CPU threads: [pulldown list] /4

In the pulldown, the only choice is 1 or 2, so I take it that 7ZFM,
like all other programs and OS, sees 4 cores but for some reason
offers to use only two.

Various programs happily see four cores here. Prime95 benchmarks
are very much along the line of what's listed in, say, Wikipedia, and
Passmark CPU scores, according to its "PerformanceTest" suit, are
nothing out of the ordinary among the tested systems with the same
CPU and OS (3236 in my case with a range of ~3100-3300 for XP
and ~3700-4000 for Win7). My Passmark RAM scores are also not
hugely out of the line among those with 4 Gb.

As suggested at some point in one of the ASUS forums, I'm tried
running the CPU with VCORE = 1.235 V (over 0.1V below default). Also
disabled C'n'Q. None made any noticeable difference so far. The system
is stable, with Prime95 running on all 4 cores continuously. (Before
you think of it: I do close it before running any performance tests )

Haven't fooled around with RAM settings yet because in BIOS they
look different from what I am used to and 3/4 of options I don't even
recognize.

Dima

 

[DK]

I installed 9.20 and ran the benchmark on it. Pretty much the same as before-
8.7MB/s and all 4 cores synchronized
within 3 or 4% of each other. It would start at ~60%, then ramp up to 100% and
stay there until the next pass. Memory
used was 851MB (forgot to mention that on the last reply).

Now THAT I find incredibly perplexing! We have very similar setups
and yet not only the test results vary widely but also a particular program
runs on them in the most utterly different ways. Crazy.

Is that possibly what you did that took 18.5 minutes?

No, I've done this test many times now with three different versions
and in all cases it was the 7z.

I will restore pristine XP SP2 image over weekend to see if it's
the OS issue in some way, shape or form. My CPUZ report is
expectedly very similar to yours save for the type of CPU and
video (I really will have to buy something; the intergated one
here totally blows - it work about as well as Radeon 7500 ten
years ago).

Dima

 

[Darklight]

I thought that AM3 offers more choices of inexpansive mobos and CPUs.
Also, I got the 840 for $86. I like keeping things frugal.


Nexus Value 430. I bought it for its inaudible fan. Has been rock solid in
the previous setup and all voltages in the current are as expected and
stable.

Dima

here read this:

http://www.techpowerup.com/reviews/AMD/Phenom_II_X4_840/

 

[Yousuf Khan]

I was following this since I upgraded from an Athlon 64 x2 4800+
(2.5GHz) on an Asus M2NPV-VM MB (4GB RAM) to a Phenom II x2 555 Black
Edition (3.2 GHz) on a M4N68T-M V2 (4GB RAM). I have noticed quite a
difference in speed, and when I used the Asus unlocker to open two more
cores, I noticed a real difference in most applications I use (some, no
change at all, but that's mostly older small-footprint apps that weren't
too slow anyhow). Windows now has it listed as a Phenom II x4 B55. In
Device Manger, open each core and update the driver. Let it go to the MS
site to get it.

You were indeed lucky to be able to unlock not just one, but two whole
cores on your X2. I had a Phenom X3 prior to this one (X6), and when I
used the Asus unlocker to get just a single additional core, it locked
up. It was indeed a bad core and not just a core locked out for
marketing purposes, it could not be unlocked.

The link that Paul provided points to a driver for XP
SP2, and is dated 2009. I'm afraid that driver would be
counterproductive on a XP SP3 system with a processor manufactured after
2009. I may be wrong, but I got better results with the MS driver (which
I normally don't use for driver updates). The "CPU driver" on the MB DVD
is nothing more than Cool 'n Quiet. If you run the setup again, it will
ask if you're sure you want to uninstall it? I left it on mine. I don't
find that it interferes with what I do.

When you use the Core Unlocker (i.e. Unleashed mode) it's been suggested
that you should not use CnQ in combination of Unleashing. You should
also disable Turbo Core (if available on your processor), and C1E
support, while unleashed.

Yousuf Khan

 

[Yousuf Khan]

Wanting more raw computing power, I went from Sempron 2200+
in ASUS Asus A7V400-MX (socket A) to Phenom II 4X 840 in Asus
M4N68T-M V2 (socket AM3).

I must say I am quite disappointed. I was hoping for about 10X speed
boost even with applications that can only use a single core. After all,
more than six years separate the two (expected boost suggested
by Moore's law would be 64X). The real life tests show nothing
of this sort! Not even close to 10X.

<snip>

Slightly on a tangent, I've been disappointed with all of my CPU
upgrades after my first one. My first processor was an 8088-10MHz on a
PC-XT clone. My first processor upgrade went from that, straight to a
386DX-25MHz! It felt like I just strapped a rocket to my machine --
everything felt faster, even the typing! Every other processor upgrade
since then has felt somewhat unworthwhile. I never noticed the
performance in day-to-day work.

Even though my current processor is probably literally at least a 1000
times faster than that first processor, they all seem to just barely be
noticeable from the previous processors, in my opinion. And I'm not one
to upgrade every year either, I usually wait about 3 years between
processor upgrades. So 386/25 was still the biggest kick in the pants
ever, even after all of this time.

Yousuf Khan

 

[SC Tom]

You were indeed lucky to be able to unlock not just one, but two whole cores on your X2. I had a Phenom X3 prior to
this one (X6), and when I used the Asus unlocker to get just a single additional core, it locked up. It was indeed a
bad core and not just a core locked out for marketing purposes, it could not be unlocked.


When you use the Core Unlocker (i.e. Unleashed mode) it's been suggested that you should not use CnQ in combination of
Unleashing. You should also disable Turbo Core (if available on your processor), and C1E support, while unleashed.

Yousuf Khan

As it turns out, I'm not running the software, but it is enabled in BIOS. Seems to work just fine that way. I don't have
Turbo mode enabled- I figure it's not going to give me that much of a boost for what I run anyway.

I haven't seen too many failures on the Phenom II x2 CPU's. I believe there were more instances of non-working extra
cores on the x4's. If I have a problem with any of them, I can always turn one or two cores off. Until that time, I'll
enjoy what I have. I'm glad I upgraded my PSU a while back since the 4 cores double the power requirement.

 

[Yousuf Khan]

In 1985, I had an IBM XT and used it to run an assembler for Motorola
6809 code. It took about 20 minutes to complete. A few years later, my
company bought us Compaq Deskpro 40 machines. BIGGGGGGGGG DIFFERENCE!
The assembler took about 30 seconds.

However, I've never experienced that kind of upgrade joy since.

Yeah, I know what you mean, things were going so fast, you almost felt
that maybe it didn't work right or something. These days, you might see
something go from 30 seconds down to 15 seconds, it'll feel like a nice
boost for a little while, but after two days you'll already be used to it.

I think it took me months to get over how much faster the new system was
over the older one.

Yousuf Khan

 

[Bob F]

In 1985, I had an IBM XT and used it to run an assembler for Motorola
6809 code. It took about 20 minutes to complete. A few years later,
my company bought us Compaq Deskpro 40 machines. BIGGGGGGGGG
DIFFERENCE! The assembler took about 30 seconds.

However, I've never experienced that kind of upgrade joy since.

My first computer was based on the intel 8008, made with hardware hauled out of
the dumpster where I worked. I then designed and wire-wrapped a processor board
for the 8080 chip. Now that was an increase!