BenchZowner
-
Posts
690 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Blogs
Posts posted by BenchZowner
-
-
Yes, be excited now
Actually I'm still in the process of finding the damned data sheet of the controller used on the reference GTX 480's, CHiL Semiconductor's CHL8266.
If you know somebody who can get the datasheet please ask him to give it to us so we can figure out all the mods.
I think I found the OCP mod, but I'm not 100% sure because I don't know if the pin configuration of the 8266 is the same as 8316's, so I'm not going to post it before I get a confirmation myself or from a brave friend
:Some good news, it looks like the CHL8266 is fully programmable via SMbus ( VID, OCP, OTP, OVP ), so if somebody with the necessary coding knowledge can pop in here we'll be able to change the Vgpu, the OCP and OTP & OVP on the fly via software.
Update 07/14/2010
No GPU OCP/OVP Firmware
You can disable OCP via software ( flash the CHL8266 controller with a modified firmware which disables OVP & OCP ) [ you still need to do the MEM OCP mod, this disables the OCP for the GPU core only ].
No need to run GPU-z and co, just flash the CHL8266 once and you're done forever
You'll find the necessary files and instructions in the linked RAR archive file:
CHiL 8266 Disabled OCP Firmware - Download from BenchZone
CHiL 8266 Disabled OCP Firmware - Download from RapidShare
Cold Slow Bug BIOS fix
Thanks to George "Hipro5" T. the Cold Slow bug fix is out in the wild.
Flash your card ( info in READ ME FIRST.TXT ) and enjoy.
nVIDIA GeForce GTX 480 Cold Slow Bug BIOS Fix
OCP hard mod *not tested yet though, the risk is yours*
original photo courtesy of TechPowerUp.com
No OCP, No OVP ( Unlocked Voltage adjustment ) BIOS & Tool
Download from: BenchZone or RapidShare
-- if a manufacturer wants to f* with somebody f* with me, come on
Summary, added 07/14/2010
To sum things up, if you want to do some extreme overclocking with a GTX 480 all you need to do is:
1) Flash the card with the ColdBug fixed BIOS ( BIOS #1 by Hipro5 ) [ download links above ]
2) Flash the CHL8266 controller with the No OCP/OVP firmware ( BIOS ) [ download links above ]
3) Do the Cold Boot Bug ( CBB ) mod as illustrated by Kinc ( mod guide pictures to be added in this post later )
4) Do the Vgpu ( GPU Core Voltage ) mod as illustrated by... numerous people ( mod guide picture to be added in this post later )
*You might have to do the MEM OCP mod, it's a hit 'n' miss thing, some people need it, some don't* ( mod guide picture to be added in this post later )
Enjoy!
-
He he thanks, that's what I'm getting lately ( the 3, you know every man's holy family
).Crappy CPU, Crappy UnCore, Crappy Efficiency, Crappy Clocks, Crappy Thermos ( lost lots of runs because I was alone and only had 2 thermos
) -
-
It is illegal.
And always use version 4.1 for HWbot.
-
Awesome CPU there!
Congrats!
-
Actually 3D Mark03 runs way better in Windows 7 x64.
wPrime is faster in Windows XP for both P55 & X58 ( unless there's something I'm missing because I tested all the OSes a few weeks ago ).
3D Mark05 is faster in Vista x86.
3D Mark Vantage is faster in Win7 x86.
AquaMark 3 @ windows 7 x64
PiFast, 3D Mark2001SE @ win xp x86
SPi32m @ winxp or win2k3server
-
There's no SMbus unless you have the Winbond chip ( which is an SMbus controller ) like the ES cards do.
Gigabyte's OC Guru "bling" Panel uses the PMbus to change the voltage.
Somebody should reverse-engineer it me thinks
-
Voltages can be adjusted via a program setting register values via the PMbus ( wish I knew how to program that
I used to be an intermediate C coder, but I'm like a beginner after 5 years of not coding at all ) -
I see a bent socket pin.
-
Say thank you Bill :-p
-
You forgot one of the do's.
Good point Frederik, rev 1.1 time
I agree with everything, except the first "don't". It's perfectly possible to create a successful competition with a "weird" benchmark, we showed that during the wprime singlecore competition:) But of course you won't get the low numbers you get with 6-core rigs etc.I partially agree with you on that.
Sure it's fun too, and stressing, but it would consume a lot of time ( wPrime 1024m on a single/dual core for example
) and some benchmarks ( like CineBench for example ) would be a matter of clock frequencies as they're not tweakable at all. -
Hey guys,
Due to the absolute failures and epic goofs that occurred in the overclocking events/competitions organized by some hardware manufacturers the last few months I decided to take some time and write an article about it, not to bash them for their silly mistakes, but to point out their mistakes and mishaps, and to give the manufacturers and the people responsible for the contests some suggestions and ideas to better organize them, leading to better competitions and therefore better products.
Whether you're an overclocker ( accomplished or not, an experienced one or a newbie ) or an employee of a hardware manufacturer, I believe this article is worth a few minutes of your time.
I'll leave you with some quotes and a link to the article.
Your comments, ideas & suggestions are welcome ( and will be added
).Back on topic now, it's time to speak about this article, what's it about and what you'll find in the following pages.The topic of this article is "Overclocking competitions" ( be it online, local or global, it doesn't matter, we'll talk about them in general and specifically as well ) organized by computer hardware manufacturers, their benefits, the nice things, and the bad things ( flaws ) and we'll try to offer some suggestions that can make these competitions better and attract more participants and... "viewers".
What made me write this article ? The fiascos of two overclocking competitions that took part online the last few months organized by two well-known manufacturers.
-Oh no, it souds like another "Bill bashing the manufacturers" article!
No no, don't worry, I'm not going to bash them ( well, I might for a bit, but I do it for them, I mean to grab their attention and get them to listen to us in order to better organize their competitions, and have a win-win scenario for both the manufacturer and us, the overclockers and the consumers ). I will try to pin-point the organizational mistakes and the ridiculous ( leading to low or even... total lack of participation ) limitations on entering the competition.
Before we go ahead and talk about the manufacturers in specific, I'd like to make myself clear and make sure that you ( the manufacturers ) don't get discouraged by our judgement and critique, we ( the overclocking community ) appreciate what you do and we just want you to listen to us and let us help you in organizing more and better events & competitions ;-)
Assuming that you are planning to host an online or local/global overclocking competition here are the key points that you want to pay attention to and follow carefully to please the overclockers, attract lots of people and have a successful contest for you and the overclockers:Let's start with the easiest part and proven to be wrong from our experience over the past few years, the DON'Ts:
DO NOT ( DON'Ts ):
* Select improper benchmarks & hardware combinations, if we're going to use a Dual-Core processor do not choose unsuitable benchmarks like CineBench, wPrime, 3D Mark06 or 3D Mark Vantage.
Wisely chosen hardware & benchmark combinations is one of the keys to a successful overclocking event/competition
* If it's a global contest with qualifiers stick to a plan, organize the qualifiers the same way for every country/region, do not handicap people by allowing E.S. CPUs/VGAs/Motherboards in a contest that people need to buy the stuff on their own to participate, supposedly you're looking for the best overclockers, not the overclockers with the highest salary.
* If it's a 2D benchmarking competition ( e.g. SuperPi, PiFast, wPrime, etc ) do not restrict the competitors to use a VGA of your own brand, limit the motherboard selection or don't limit the hardware at all.
If it's a 3D benchmarking competition ( e.g. 3D Mark, AquaMark 3, etc ) do not restrict the competitors to use a motherboard of your own brand, limit the graphics card selection or don't apply any brand limitations at all.
* Don't make the competition more unfair than it really is by sending cherry-picked samples to a few people or just one ( considering that in a competition where the participants need to buy the hardware on their own the overclockers with a higher budget can buy more pieces in order to find a better CPU, motherboard and/or graphics card, things are unfair already, no need to make things even harder ).
* Do not organize a product launch contest when the product hasn't been launched already or it was just released and its retail availability is really limited, give the people some time to get a chance to buy the product and get ready for the contest....
Here's the link to the full article ( click me! )
Enjoy!
-
Windows XP Professional x86.
-
I don't know if it was hicookie posting scores via multiple accounts.
But from my communication with a few Gigabyte contacts that I have I have the impression that they were working under pressure and they needed to attract as many people they could to participate in this contest, there was an "urge" to get people to submit scores.
Maybe an angry boss ? I don't know, but that's how I comprehended their e-mails and behavior.
-
Yeah, how much for the best retail chip ?
-
I'm 98% sure that it's a change from the engineers that made the "old" software not to function properly.
The Vddci, Vpci-express & Vmem work just fine, so unless I'm mistaken and my info is bad about Gigabyte fixing this, this shall be fixed soon with a new version of the utility by Gigabyte
-
2 CPUs, one of them being Stummie's and the other one being... well... you know tsan and his luck isn't exactly what I'd call picking up, not even close yet.
If anybody else finds a retail as good as tsan's seems to be, I'll be surprised.
-
You guys are either bad overclockers or just went through a batch Intel planned to send to me initially
-
I might be a tad late to the party, but I ensure you that it definitely is read-worthy for everyone, the average joe, the budget overclocker and well... the extreme overclocker too.
I just finished and published the review of the newest mainstream Dual-Core processor from Intel, the Core i3 540.
Some quotes from the review:
For the past one and a half ( or so ) year Intel had nothing new to compete with the Phenom II and Athlon II processors from AMD except the "old" Penryn processor family ( Core 2 Quad and Core 2 Extreme QX9650 / QX9770 ) whose pricing wasn't exactly what I'd call competitive. If you were after a cheap processor ( 99$ - 199$ ) your best bet was AMD's Phenom II. It took Intel 1 year to release a Nehalem-based entry level & mainstream processor family ( I'm quite sure they had the processors ready for a while, but don't know why they didn't release them earlier ).Now, with the release of the Core i3 and Core i5 processors, Intel's processor lineup for 2010-2011 is complete, we have the very cheap Intel Core i3 dual-core processors, the relatively cheap Intel Core i5 dual-core & quad-core processors, and finally the high end Intel Core i7 quad-core & six-core processors.
Today we're going to have a good look at the affordable entry-level processor from the Core i3 lineup, the i3 540.
The Core i3 540 is a dual-core processor with HyperThreading support based on the Clarkdale core ( which is based on the Westmere core ).
It's the first CPU with an integrated GPU from Intel for the desktop processors segment, although the IGP ( as Intel calls the integrated GPU ) [ Integrated Graphics Processor ] isn't actually on-die, it's on a separate die installed on the same package with the Clarkdale core.
Just like the Core i5 7xx CPUs, Clarkdale also features an Integrated PCI-Express BUS controller and a DMI controller, it's like a dual-core i5 7xx with an extra component ( Integrated GPU ), a different memory controller, manufactured with Intel's 32nm process.
Currently there are two Intel Core i3 processor models, the i3 530 and the 540, with the only difference between them being the clock frequency ( 2.93GHz & 3.06GHz respectively ) and the price ( MSRP: 113$ & 133$ respectively ).
If you asked me to guess if you're bored already or not, I'd say yes. I'm not good at all in writing introductions, so I'm going to do you a favor and end this suffering by letting you know what you'll find in the following pages of this review.
We're going to talk a bit about the technical specifications and the Clarkdale architecture, take a look at the CPU and the boxed heatsink, test the processor at its stock clock frequency and overclocked, compare it with an i7 950 processor in various configurations ( stock, clock per clock at 4GHz with 2 cores disabled to match the i3 540 in core & thread count, at 4GHz with 2 cores & dual channel memory, and fully overclocked ), find its max limits with the stock cooler and an aftermarket CPU air cooling solution, and then do some extreme overclocking using Dry Ice & Liquid Nitrogen.
If all of the above excite you or just seem to catch your interest, grab a cup of coffee or a cold beer and move on to the second page of this review.
Ok, now I'm going to try to explain to you what I wanted to check, how I did it, and why I did it to give you a general idea behind the methodology I used for this review.Usually when I'm planning the testing for a review I try to cover every aspect of the test subject ( the product, duh! ) and take the appropriate measurements to compare it with various hardware from the same manufacturer and the competitors offerings, test its raw and everyday performance in various tests & benchmarks, and finally, push it to its limits ( or even beyond that :-p ).
My first idea for this review ( no,no, testing the product at its stock settings isn't an idea! do you think that you have to breathe ? ) was to compare the performance of the Core i3 540 clock-per-clock ( at the same clock frequencies and core & threads count ) with one of the 45nm parts of the previous generation to see if things got any better at the same clocks. So I overclocked the Core i3 540 at 4GHz and run the benchmarks, and also did the same with the Core i7 950 at 4GHz with only 2 DDR3 DIMMs ( to have the same amount of memory and of course to run the IMC at Dual Channel memory mode ) & 2 cores disabled from the BIOS setup to match the core & threads count of the Core i3 540 ( 2 cores / 4 threads ).
Since I had everything set up for a clock-per-clock comparison to see if the new 32nm parts have any performance advantages at the same clock frequencies I said "Why don't I test the i7 950 at the same settings but with 3 DDR DIMMs and the IMC in Triple Channel mode to see if there's anything to be gained by the increased memory bandwidth from operating in triple channel mode ?", and so I did. I installed a 3rd DDR3 DIMM and re-run all the benchmarks.
And the sweetest part of all, overclock all the CPUs to their maximum stable clock frequencies and run the benchmarks again, watch the scores go sky high ( well, for air cooling I'd say they did great, for the absolute clocks I need to get my liquid nitrogen dewars filled and overclock them at "stone cold" temperatures, way under 32*F/0*C).
To read the full review, just click here!
The Extreme Overclocking page should be filled with contents this weekend, I'm getting ready to test the i3 540 and an i5 670 on LN2 ( might have a live stream as well )
-
Its all down to luck
There WILL be a retail that can match ES in the future, look at the 975 D0
BZ sorry for being arrogant earlier, I get a little worked up sometimes
*Shake hands?*
Myles
We're ok man
We all get a bit over our heads sometimes, it just happens
-
Not too shabby.
Wish I could find at least a similar chip
-
I certainly hope to see some good batches or a better revision on the shelves, but I can't see that happening soon.
Unless the shops start to throw 980X's to the garbage bins, for the time being I can only find 3003B chips.
-
I don't know about the rest ( and I'm fine with ESs being allowed as long as the retails can hit similar frequencies ) but I'm fine with people handpicking processors.
-
The same can be said about the Core 2 Extreme X6800 and their ES's, the retails were better, only difference being it wasn't illegal or that much of a risk to buy a perfectly legal and backed by a warranty chip
What's unique in this case is... the gap that separates the retail and the ES processors this time.
With the 670 ES & retails ( where the ESs are once again better ) the difference isn't that big, yet if you want a top spot you need an ES again.
But the gap between the 980X retail and 980X ESs is abysmal.
We're talking about several hundreds of MHz.
Pi 1M ? 6745MHz vs 6153MHz
Pi 32M ? 6587MHz vs 6091MHz
3D Mark06 ? 6358MHz vs 6089MHz
3D Mark Vantage ? 6090MHz vs 5700ish MHz
AquaMark 3 ? 6485MHz vs 6161MHz
NVIDIA GTX 480 Voltage Modifications
in Nvidia GPU Overclocking
Posted
OCP hard mod & disclaimer added.