CPU upgrade, from T6400 to X9100. Because 3 GHz are way better than 2 GHz…
UPDATE: After a few months the CPU upgrade turned to be a remarkable failure. I advise anyone against this kind of practice and I invite you to read the post regarding my useless troubleshooting efforts.
I purchased my latest computer in absolute emergency conditions, and except for an annoying, sound-related issue when I extensively use the network (a fact for which I would be inclined to blame and damn Vista SP1) I’m satisfied with it until now. But being obliged to spend a limited budget obviously didn’t hinder me to upgrade the system main component, the CPU, overlapping to satisfaction the pleasure of having a fairly recent setup to let me use it in scenarios that are a little less retrograde than the ones I’m usually accustomed to.
The laptop Acer Aspire 6935G I purchased in November featured an Intel T6400 Core 2 Duo 64-bit processor with a 2.00 GHz clock, 2 MB level 2 cache and 800 MHz FSB. Nowadays a 2 GHz clock is below the minimum required for a decent computing experience, and in my case it quickly showed to be inadequate. Doing the proper research I learned that the platform on which the Aspire 6935G (G means discrete Graphic component) is based, the PM45 chipset also known by the “Cantiga” codename, has the QX9300@2.53 GHz Quad Core and the X9100@3.06 GHz Dual Core at its performance top.
Being at the moment unsure if I could adequately exploit the parallel computing capabilities offered by a Quad Core, I’ve convinced myself that an X9100 Core 2 Duo would have been the ideal upgrade to push my laptop’s performances. After having checked compatibility between the new component and the chipset, on January I purchased the CPU on eBay for 249 pounds/284.31 euros (free shipping from the United Kingdom) and soon after I installed it on the PC with the aid of this short visual guide found on-line.
In the end the guide was useless because the entire upgrade process is rather straightforward for someone accustomed to play with hardware since many years, and it didn’t help me anyway when at the first start-up after the upgrade the screen didn’t show any sign of life. The problem dragged itself for an entire weekend I spent to remove the X9100 and reinstall the T6400 (useless, the screen was still dead), update the firmware to the latest available version and connect the laptop to an external CRT screen (the new firmware didn’t have any effect, and the external monitor confirmed that the LCD was the problem), remove and reinstall everything a dozen times just to be sure I didn’t do anything wrong.
The issue was resolved in the most classical way, i.e. when I changed perspective by shifting my attentions from the laptop’s back to the front and finally finding the problem in the screen power cable hidden behind the panel covering the speakers, which I don’t know how I disconnected while I was installing the CPU on the back. The result of all this hard work is depicted by the following screenshots: the old T6400 Dual Core leaved place to the new X9100 “Extreme”-class Core 2 Duo at 3.06 GHz, equipped with 6 MB second lever cache, 1066 MHz Front Side Bus (the RAM is made up by two DDR3 banks manufactured by Micron, I’ll skip over the timings details), support to Intel VT-x virtualization technology (I’ve got the hypervisor! finally I’ve got the hypervisor :-D) and an unlocked clock multiplier for overclocking the CPU frequency.
The cold numbers
Once I installed the hardware it came the time to prove the X9100 with software. Aside from verifying the upgrade effects with everyday use and the single applications that could benefit by more power - ie video games - I’ve used some benchmarks freely available on-line. Except for Street Fighter IV, which has been useful both as a benchmark (with unusual and largely negligible results) and as a separately-taken application. Every test is here depicted with a results summary, explanations and screenshots.
Euler3d Benchmark 2.2
Euler3d Benchmark is a test provided by Computational AeroServoElasticity (CASE) Laboratory at the Oklahoma State University. It’s purpose is to “provide information about the relative speed of different processor, operating system, and compiler combinations for a multi-threaded, floating point, computationally intensive computational fluid dynamics code“. Once started with its own script, Euler3d skips the output files and the results visualisation of its complex calculations to simply provide the elapsed time and a score defined in Hz.
A rather boring fluid dynamics benchmark
It is possible to define the maximum number of steps to use for every test, and in my case I have chosen to use the default value that is 5. The benchmark has been executed for three times in a row after the system start-up, finally the three results have been combined and divided for 3 to obtain the following average values:
As you can see the gain scored by the X9100 CPU on the old T6400 Core 2 Duo is almost 40%, a value that is clearly bigger than the 1 GHz clock difference and that in fact also depends on the increased RAM frequencies (1066 MHz against the previous 800 MHz) and the particularly wide L2 cache included in the “Extreme” Core 2 Duo CPUs (6 MBs against the previous 2).
x264 HD Benchmark 3.0
With x264 HD Benchmark we shift from fluid dynamics calculations to encoding a 720p high definition video clip. The benchmark uses a series of small tools and scripting languages to automate the included clip conversion in the x264 HD format. The benchmark converts the video in 2 passes, with 4 series for every pass hence a total of 8 passes.
Another boring textual benchmark
The benchmark has been executed once after the system start-up, and for every pass I’ve calculated the frames-per-second average value obtained after the 4 series with the following results:
In this case the performance gain of the new CPU ranges between 34% (first pass) and 37% (second pass), a value that tends to confirm the results of Euler3d benchmark and the positive effects of the new system configuration due to the X9100 Core 2 Duo.
CINEBENCH R10
CINEBENCH’s application field is 3D animation and rendering. Based on the renowned animation software CINEMA 4D (employed in films like Beowulf, Spider-Man 3, The Golden Compass and The Polar Express), the CINEBENCH release 10 used for the test measures the processor speed by calculating the time elapsed for rendering a single 3D frame at 800×600 pixels (using one or both the CPU cores) and the GPU performances with the animation of a fixed scene.
The 3D frame rendered by CINEBENCH R10
In the CINEBENCH case too I’ve made three tests after the start-up, including the single-frame rendering with one or two processor cores and the OpenGL animation with the GPU. These are the average results:
CINEBENCH confirms the usual trend of performance increase ranging from 30% (+36.76% in the single core rendering) to 40% (+39.4% in the multi-core rendering), with a lower improvement in the 3D animation. The GPU profits by the 3.06 GHz X9100 processor marking a +24.55% performance gain, even though I have to say that by doing the test again with updated graphic drivers the GPU score still improves up to 4755 points, with an over 30% improvement on the previous tests.
Street Fighter IV
The new chapter of the esteemed beat’em up saga by Capcom features an internal benchmark with hard-coded parameters: everything the user can do is to launch the tool, watch the precalculated fights and the passing through the several post-processing effects (unique to the game’s PC engine) until the final results are shown. Even this way, however, the test is interesting because it measures the system power in a real world usage scenario.
Finally some action
The benchmark has been executed for three times in a row after the system start-up, the “rank” the game awarded to my system has always been the same (”C”) both with the T600 and the X9100 and these are the average results:
Street Fighter IV seems to run paradoxically better (even if only slightly) with processor and memory working at very lower speeds! Theoretically the almost non-existent difference between the results would suggest that the game behaves identically both before, and after the CPU upgrade. Actually this is only partly true, because if the pure 3D graphics performances remain the same, with the new processor I have been able to permanently activate some visual options that heavily hit the action smoothness before. I will return to the matter soon.
EVEREST 5.30
EVEREST Ultimate Edition, the diagnostics and benchmarking tool made by Lavalys offers some synthetic tests that, quoting the program’s manual, “show only the theoretical (maximum) performance of the system” providing “a quick and easy comparison between computer states“. The entire EVEREST benchmarks set measures memory read, write and copy speed, the CPU performances while dealing with different computational algorithms (Zlib compression, branch prediction, AES encryption) and calculation of several frames belonging to popular fractal types (Mandelbrot, Julia etc.) by the FPU.
Even considering the fact that these are synthetic tests, the difference existing after the processor upgrade is widely highlighted by the score progression with margins ranging from the minimum of memory copy speed (+11.06%) to the maximum of PhotoWorxx benchmark on the basic tasks (arithmetic computation, multiplication and memory subsystem) employed in digital photo processing (+48.81%).
Daily use and conclusions
Benchmarks are certainly useful to verify, with meaningful numeric values that are comprehensible to external observers, the performance improvement due to switching to a higher-class processor. What however depicts the performance boost in a more effective manner is the everyday experience, opening the usual applications and utility software, Internet, games, multimedia. In the everyday practice an upgraded system shows both the best it can do and the worst as well, because more power means (especially on a laptop) that you must learn to deal with some drawbacks.
Let’s talk about the positive sides first: with the upgrade to the X9100 CPU, the application software I use the most have experienced a fair yet noticeable performance improvement. Word and Excel load and run even more faster and smoother than they did with the T6400 and the same can be said for ACDSee and Ulead GIF Animator as well; Firefox and Thunderbird have very fewer uncertainties when I load tens of tabs at the same time or I view some e-mail full of images and nasty HTML code (damn Play-Asia.com newsletter…).
If more demanding multimedia contents didn’t pose particular problems to the processor even before the upgrade (Adobe Flash aside), now the CPU usage has dropped below the perception threshold (10-30%) with high definition films at 1080p too (Blu-ray disks or Matroska rips it’s the same) granted I use VLC media player or PowerDVD. The operating system itself (Windows Vista SP1) is more reactive and lively, Aero GUI graphic effects are smooth and generally without faltering, loading the system components (from Event Viewer to Windows Media Center going through the Control Panel) is an essentially immediate affair.
The current system performance index
The impact of the new CPU on the games that were on the system before and short after the upgrade has been variable: Valve’s Portal remained identically enjoyable even with the addition of another 1.06 GHz and a faster memory; Serious Sam HD: The First Encounter identified the X9100 CPU with a higher performance level (”High” instead of “Medium”) in the advanced options but the GPU still is the bottleneck in real-time rendering; Braid was never particularly demanding over hardware resources, but the slight hitch the game seemed to show (and I repeat: it seemed to show) in its first moments (when the “hero” goes from the dark of the city roofs to the residence entrance) should be vanished; Capcom’s Resident Evil, a conversion from PS1 designed to run on Windows 95, must be slowed down to 30% of the processor speed (Mo’Slo courtesy) instead of 50% as before.
Contrariwise to what highlighted by the game’s internal banchmark, the new CPU provides Street Fighter IV the power needed to flawlessly manage the graphic post-processing effects exclusive to the PC version (”Ink”, “Watercolor” and “Posterize”). Before the upgrade their activation marked a noticeable hit on the game smoothness while now, from what I have seen, the 3D graphics goes to the max in 99.99% of the cases.
But where the upgrade effects impact the most, and the fact doesn’t amaze at all, is for the emulation of relatively recent video gaming platforms: the old MAME32 version I still use (0.90) is now capable of emulating Fighting Layer (a 3D arcade from 1998) at the maximum speed almost all the time, and some PS2 games that previously limped a lot on PCSX2 now run hugely better (Maximo is even faster than before, Saint Seiya is really enjoyable, Ico would run very well if graphics wasn’t still so messed up while the 2D beatm’em up Hokuto No Ken runs almost always to the max).
And let’s talk, in the end, about the upgrade drawbacks, the negative sides that balance the greater power and computing capability highlighted in the post: the first, plain difference between having a 2.00 GHz processor and a 3.06 GHz one is the drastic rise in power consumption, which because of the maximum TDP increase (from 35 to 44 W) drains the laptop’s battery very fast once the PC is unplugged from the power grid. From the more than two hours of the T6400, in normal use conditions (websurfing, Explorer usage and a little more) and high performances power saving profile, the battery duration is reduced to less than half and an hour. And it goes below an hour if I load some 3D video game or a HD video clip.
Another pretty “visible” drawback is the overwork the fan connected to the passive CPU-GPU cooling system has to bear. Now the poor thing makes a dreadful noise and always goes crazy when I load PCSX2, while before its presence was discreet and almost never noticeable. After a bit you become accustomed to that noise, that’s true, but there still remains the nuisance of distinctly hearing the fan spins (and decibels) increase in a system that would be much more quiet otherwise.
Related posts
- Emulation galore: news for ScummVM, ResidualVM, PCSX2 and Supermodel!
- The aftermath of a failed upgrade
- 3dfx Voodoo graphic cards emulation coming to DOSBox
- Snapshots #3: high definition, high emulation
- Finally the upgrade to Firefox 3(.5.2)
- PCSX2 0.9.6, PC emulates the PS2 way better than PlayStation 3
- Parallel computing and GPGPU, the super-PC genesis between universal libraries and proprietary platforms
- Intel, AMD and NVIDIA at war for high performances graphics cards
- Upgrade to Firefox 3? Sure, but quietly
Comments
15 Responses to “CPU upgrade, from T6400 to X9100. Because 3 GHz are way better than 2 GHz…”
Leave a Reply
-
· Subscribe to Sir Arthur's Den ·
-
· © Some Rights Reserved ·
The exclusive contents of Sir Arthur's Den are distributed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License, and must be ascribed to Alfonso Maruccia as the sole owner of the related copyrights.
Hey there!
I read your article because I am in the same situation with the T6400 that I want to upgrade. I noticed you choose the X9100 CPU and that you are complaining about the noise and battery the new CPU takes, since it is 44W (45?). I just have a simple question:
Why did you not try the T9900? I boasts about the same price, has the same speed but is only 35W?
Perhaps I missed something?
Cheers!
/Mehdi
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.1; sv-SE; rv:1.9.2.8) Gecko/20100722 Firefox/3.6.8
If I recall well, I chose the X9100 for the overclocking feats it has (unblocked multiplier).
The problem is: I’m still waiting for some decent update (or modification) for the crappy, crappy UEFI-BIOS my Acer Aspire has (ie I can’t change a fucking thing right now not even set the hardware virtualization on!!!) to exploit it.
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.0; it; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3
hello
i have acer aspire 6935 and i was tried to use c2q q9100 but doesnt work becouse of bios.
are you sure that on this laptop qx9300 will work?
Browser
Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.1; WOW64; Trident/4.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; Media Center PC 6.0; .NET4.0C; InfoPath.2)
It should work, yes. The chipset/socket should support it.
Anyway I suggest you not to touch your laptop’s CPU anytime soon: eventually the upgrade went bad and I was forced to switch back to the T6400 C2D.
I’ll address the issue with my next post, it should be up in the upcoming days so you can subscribe my feed to know what went wrong and why….
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.0; it; rv:1.9.2.8) Gecko/20100722 Firefox/3.6.8
Hey there! Wanted to update you guys with the upgrade that I finally did
I upgraded from the T6400 to the T9900 3.0Ghz and I’ve been using it for almost 6 months now!
The biggest difference that I noticed immediatly was that Windows works shitload faster and the apps start faster, the computer definetly feelt faster and I even did a few speed tests with converting and so on. To be honest, the difference wasn’t like day and night, but it sure felt good having a much more speedy computer. And I do not regret for not choosing a quad-cpu instead. I honestly haven’t noticed any difference in the fans running at higher speeds or that the battery runs out quicker, it feels about the same.
My tip is, IF you have a 2.0Ghz or slower, make the change to the T9900 and you WILL feel the difference, but otherwise it’s not worth the difference if you have faster than 2.0 Ghz.
A friend of mine bought a new computer with the “slowest” version of i7 mobile, and he paid about 50% more than what I did with the total price including the new CPU, and I’ve been testing his computer a little and it “feels” just like my computer when it comes to watching movies, surfing and listening to music, which both of us do a lot.
What this simply means is that C2D T9900 is still the shit!
Cheers
/Mehdi
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.1; sv-SE; rv:1.9.2.8) Gecko/20100722 Firefox/3.6.8
Hey guys,
Awesome to read all these successful attempts.
I really want to do the T6400 to T9900 upgrade, but im not too sure how to do the upgrade. Dont really have experience with laptop upgrades. I have done desktops though. Anyway is there anything i need to watch out for? How do i proceed this operation exactly?
Thanks alot guys
Peace.
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.13) Gecko/20101203 Firefox/3.6.13 ( .NET CLR 3.5.30729)
My advice is: don’t do it. I hope to blog about this in the upcoming days….
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.0; it; rv:1.9.2.8) Gecko/20101203 Firefox/3.6.13
Hello ! I have an Acer Aspire 8735G with CPU T9600 and I want to upgrade it to T9900. What about X9100 or even QX9300 ? Still, does it worth ?
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.8) Gecko/20100722 Firefox/3.6.8
No, it isn’t worth at all. If you will follow the blog’s feed, I’ll explain why within the next days….
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.0; it; rv:1.9.2.16) Gecko/20110319 BTRS35926 Firefox/3.6.16
Heyho,
just want to leave you a reply about the X9100. I have one in my 14″ Notebook too. All you need to use it, is Throttlestop. This tiny little tool lets you control the multipliers and also gives you the opportunity to select the VID for the desired speed.
Very important for battery use, because the X9100 has some disadvantages compared to the T9900. One is that it doesnt power down as deep as the T9900 and that it doesnt support VIDs lower than 1V.
If you know this there is no more disadvantage. Have fun!
Browser
Opera/9.80 (Windows NT 6.1; U; de) Presto/2.10.229 Version/11.60
Hi guys!
A new update from my old trusty T9900!
I’ve noticed that yeah, my old trusty T9900 was getting a bit “slow” and I had a friend with a i7 portable that was faster than mine and I was thinkin’ about selling the old workhorse and getting a faster portable also….until….enter SSD!
Bought the Intel 320 model, and holy crap, my trusty old T9900 is on steroids!
If you’ve got a trusty Core2Duo and you’re thinkin’ buying new, just plug in an SSD instead, it works very fine with my Acer, since mine is a big model it has 2 slots for SSD (120Gb for OS) and an trusty HD (1TB) for all my music and films.
Seriously, it runs so much smoother than I think I’m gonna still use my T9900 for at least another year since it runs all programs well!
Anyone else who has made this upgrade? What are your thoughts?
I’m not saying the programs “work” faster, but really, it is a big difference when the OS runs boots much faster and programs start much quicker.
Browser
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:8.0) Gecko/20100101 Firefox/8.0
Nope, I don’t like SSD technology (too much unreliable for my tastes) and I don’t value data transfer speed over every other thing.
What I need is more computational power (CPU and GPU), and as you can see from the update to this post I’m convinced I’ll effectively need a new computer (be it laptop or rather a desktop machine) to get it.
End of line
Browser
Mozilla/5.0 (Windows; U; Windows NT 6.0; it; rv:1.9.2.16) Gecko/20110319 BTRS35926 Firefox/3.6.16
I have been reading the posts. My current situation is that I purchased a q9100 to go into my toshiba l5o5, which has a T6400. I get a black screen when I boot (and no its not the screen cable!!!), I have looked everywhere online and found that it should be fully compatible, especially since both are penryns. Toshiba us does not have a bios update, and my laptop won’t take the UK updates. furthermore, under comparison the two chips are slightly different as far as 45w to 35w, and the Q9100 has no parity bit. Does anyone know the fix to the problem? By the way the windowa to the drivers, and I can see the screen in safemode. what is the fix im missing here?
Browser
Mozilla/5.0 (Windows NT 6.1; rv:9.0.1) Gecko/20100101 Firefox/9.0.1
Has anyone read the last reply?
I intend to put a cooling system so Im not worried about the heating problem. and I will try the throttlestop program, but right now Im in ditches with this processor. I know that windows took the drivers I can see it through the back door. So something else is preventing the GM45 from operation. I have pursued motherboard jumpers as a fix to the 800Mhz to 1067Mhz FSB change and found no resolve there so I know its BIOS related. I would run Clockgen as an ultimatum, but am not yet convinced of its capabilities.
The BIOS Toshiba provided is proprietary at best and doesn’t have great functionality as the desktop Pheonix-Award BIOSs’ have…
So, if anyone knows the answer to this challenge it would be greatly appreciated. Even if it is a new BIOS download I could use? Im pretty positive that the Parity bit selection’ has something to do with it?
Awaiting comment…
Browser
Mozilla/5.0 (Windows NT 6.1; rv:9.0.1) Gecko/20100101 Firefox/9.0.1
By the way-
The signaling logic to the clocking IC is
FSA FSB FSC
667Mhz 1 1 0
800Mhz 0 1 0
1067Mhz 0 0 0
,but no jumper is installed on the main board, so this would be a firmware related issue!
Browser
Mozilla/5.0 (Windows NT 6.1; rv:9.0.1) Gecko/20100101 Firefox/9.0.1