[Person of Color]

I have so many high flow fans now I'm not even worried.

[DustFalcon85]

 Viper The Rapper, on 11 October 2013 - 01:06 PM, said:

If you live near a Micro Center, return the current board and get an i5/mobo bundle. Or an AMD bundle. Your choice. No one beats Micro Center on price. Not even Shady Pauly's, and that guy is pretty damn cheap.

"Get an i5-3570K and a motherboard that is capable w/ overclocking and OC that bitch to 4.5 GHz. Even Randy Pitchford can do it" Just like Viper said, Forge.

[Person of Color]

You're starting to creep me out.

[Forge]

I really don't have the funds to back out and get a new board & cpu
there's nothing wrong with the board, there's nothing wrong with the chip. they just don't get along very well together. probably why most new uefi windows 8 compliant boards don't have core unlocking features

when it's time to upgrade again in five years i'll see what has more bang for buck in my price range (this time around it was AMD)

[Person of Color]

I've given up on this new board. The CPU frequencies dance around like crazy when it tries to reach 4.0. It's not possible to get it stable at 4.0.

[Jimmy]

 DustFalcon85, on 17 October 2013 - 05:56 AM, said:

"Get an i5-3570K and a motherboard that is capable w/ overclocking and OC that bitch to 4.5 GHz. Even Randy Pitchford can do it" Just like Viper said, Forge.

[Forge]

 Viper The Rapper, on 18 October 2013 - 07:34 AM, said:

I've given up on this new board. The CPU frequencies dance around like crazy when it tries to reach 4.0. It's not possible to get it stable at 4.0.

that sucks

is it the shitty vrm, or is there something else throttling it?

i had the rev 4.0 board at a "stable" 4010 with the 560 phenom, but the random freezing due to having the cores unlocked was too annoying to deal with. I locked the cores back up and i'm running the dual at 4090.

[Person of Color]

It's not throttling, it just can't maintain a stable clock speed.

The bus will vary by at most 1MHz, but the CPU clock is all over the place. It will surge from 4010 to 4066/4081 randomly and crash the machine. It's not possible to get a stable clock speed.

I'm most likely going to be selling my CPU/Heatsink in a couple weeks. I'm probably going to exchange this board for an 1150 board and grab a Haswell i5.

[Person of Color]

LOL all these people on Overclock.net are accusing me of not knowing what I'm doing and trying to convince me to stay in the AMD fold.

I've overclocked 486's and S3 ViRGE's. Fuck these people. You ask for advice and it becomes a dick size contest. Not that they'd know how to use it.

[serpe7]

 Viper The Rapper, on 18 October 2013 - 07:34 AM, said:

I've given up on this new board. The CPU frequencies dance around like crazy when it tries to reach 4.0. It's not possible to get it stable at 4.0.

You have to enable HPC mode in bios at the Cpu Core Features and then set LLC to Extreme in order to get a stable frequency/voltage
under cpu full-load. Also, disable Core Perfomance Boost. You must do all these 3 changes, otherwise you'll get throttling.

Then, try to find the correct vcore by increasing or undervolting the offset +-, always test in prime until a core gets an error.

[Person of Color]

I don't have HPC mode or Core Performance boost. I'm running a Phenom, not a Piledriver.

What I have to do is sell this CPU and get a 4670k.

[serpe7]

If i remember correctly HPC appears in the latest bios for 970A-UD3, FC. If you have FB try updating the bios, unless you meant the feature is there but unavailable.

I have the FX-6300 and only with this feature enabled I got a stable frequency under full-load, don't know about the phenom though.

Anyway if it doesn't work you can disable APM manually with AMD Overdrive. Go to Core/Voltage, enable Turbo Core, then disable it, this way apm is disabled and you get a stable frequency. But HPC worked for me, so I don't have to set it manually anymore with amd overdrive.

[Person of Color]

I'm running FC. The Phenom II lacks these features, only the ShitDozers have them.

[Forge]

 serpe7, on 21 October 2013 - 09:54 PM, said:

If i remember correctly HPC appears in the latest bios for 970A-UD3, FC..........etc

the uefi is not a "static" bios. if the chip does not have feature X it will not appear in the bios menu

as much as the board manufacturers and AMD spout backwards compatibility; that went out the door when windows 8 was released and they all geared their products to cater to it. New boards cripple phenom chips. They're usable, but you can no longer get the full potential out of them.

[Person of Color]

Looks like this whole shitty saga has ended.

I bought a used i5 2500k off eBay. Unfortunately, it's the world's shittiest Sandy Bridge, so It needed 1.428v to hit 4.5GHz! I shit you not. It also needed a PLL voltage of 1.9 and CPU PLL Overvoltage enabled. I coupled it with a Hyper 212 Evo with a second Cougar silent 50cfm fan in push/pull. Pretty quiet and takes half an hour of Prime to peak out at 62 degrees!

I coupled it with a Gigabyte Z77-D3H motherboard. I really like this board. It's an absolute bargain. It's pretty good for overclocking. Not great, but pretty good. I think the LLC is too aggressive, it was safer for me to run the processor at 4.28 than have a higher LLC setting compensate for 1.416v. Anything above "medium" would just overvolt the shit out of it. But for a $117 motherboard it's the tits!

http://www.gigabyte....spx?pid=4720#ov

All in all, I was hoping for 4.7/4.8GHz but I can't complain. This thing is incredible as is. I only paid $165 for the chip, $50 for the cooling, and between Staples refunding me $35, and the future sale of the Phenom II and old heatsink, the whole upgrade only cost me $85!

[Forge]

Thanks to your recommendations I'm happy with what I got.
GA-UD3-990FXA (Rev. 4) $100 U.S.
FX 8320 $150 US
also using a Hyper 212 + in a push-pull dual fan format with a max load temp of 55.1c after several hours of prime95

i still haven't got around to messing with the default volts, but as it sits now it's still a nice 24/7 setup (good processing power with low heat)
i also haven't dug around in the bios too deep yet either so i need to find the APM setting and disable it to keep it from idling down to a lower Mhz when not in use

[Person of Color]

Wow. Nice overclock! The FX series are going to get faster with time. Their somewhat weak spot is gaming, and the next gen consoles are going to have 8 weak x64 CPU cores, forcing developers to optimize their software. The FX chips will REALLY benefit from this. Those things haul ass with proper multithreading.

[Person of Color]

Hey Forge what graphics card are you running?

[Forge]

Zotac Geforce GTX 650 Ti Amp

it was the best i could afford. the main weakness is the 128bit bus speed. A 192 would have been better, but it was out of my price range at the time (probably still is, but i stopped looking after i purchased this)

[Person of Color]

650's are for poor people and should be eradicated. Everyone knows Mikko's 780 Ti Quad SLI setup is the best. He keeps the receipts under his economics degree.

[Forge]

 Viper The Rapper, on 08 November 2013 - 11:20 AM, said:

650's are for poor people

this

 Viper The Rapper, on 08 November 2013 - 11:20 AM, said:

Everyone knows Mikko's 780 Ti Quad SLI setup is the best. He keeps the receipts under his economics degree.

i can still play most polymer maps at a decent framerate so i'm not too envious

[Person of Color]

Polymer runs like shit on my card because Plagman sucks at programming. It's the only software that's given me headaches after being a Radeon owner for two whole years.

He's part of a Jewluminatti conspiracy to sell more nVidia cards.

[The Commander]

Oh shit...

[Hendricks266]

 Viper The Rapper, on 08 November 2013 - 11:47 AM, said:

Polymer runs like shit on my card because Plagman sucks at programming. It's the only software that's given me headaches after being a Radeon owner for two whole years.

Try running more open-source OpenGL-intensive applications. The only reason commercial software runs on Radeons without apparent bugs is because the companies that make the software pour big bucks into having AMD's shitty drivers R&D'd against their product.

[Person of Color]

I do. Nothing else gives me issues. Why is it Nvidia owners bitch about ATI drivers more than we do?

[The Commander]

To be honest, Polymer was the only thing that ran like crap on my old ATI, and I run lots of shit...

[Person of Color]

I don't like polymer to begin with, but if I were writing a piece of software, I would want everyone to be able to experience my handiwork, not just people who own products my employer makes.

[Helixhorned]

The thing is, when you don't own actual hardware of the "other products", then you simply don't realize that they may be behaving differently from what you experience on your own machine. A lot of behavior is explicitly specified by the OpenGL Standard(s), but as with many pieces of technical specification, it leaves various details to the implementors. So ATI and NVidia cards may be behaving differently in some aspects, but as an application programmer you can biased into thinking that the behavior you're experiencing is the "correct" one and won't notice that it's not unless you actually test it out on a variety of hardware.

As a concrete example, it seems that you can draw a scene "piece-wise" on NVidia cards, like this:

(At the end of each bullet, the front and back buffers are to be swapped.)
- clear screen
- draw, say, one tile in a rectangular tile arrangement
- draw another one...

This one actually worked perfectly fine in Mapster32 until people came complaining about the scene flickering. The resolution of this riddle here is that after a buffer swap, OpenGL leaves the contents of the back buffer completely undefined but the NVidia implementation apparently chose (for convenience, or maybe because there was no overhead in doing that) to behave as if with each swap, the contents of the front buffer were transferred to the back buffer.

The bottom line is, both implementations behave correctly as far as the OpenGL spec is concerned, but my unawareness of this point led to the mentioned Mapster32 bug for owners of ATI cards. (It was fixed in r2115.) So, instead of complaining aimlessly, one would be better off providing experimental data that may be useful in identifying these cases of implementation-specified behavior, by say, doing comparisons of one particular scene on different systems.

Then again... Plagman has repeatedly stated that Polymer is in need of redesign and he knows what needs to be done. So my rant is simply about dispersing the simplistic "one company is better than the other" view.

[Forge]

i'm not bashing one company or the other (both have good and bad aspects, i've owned one or the other at various times and i was satisfied with both), but it seems a bit silly to exclude a major company's hardware from being able to use a bit of software properly

-the flip side of course being that it's free and worked on in people's free time - but this leads to the degradation into the realm of stupidity - "ATI is crap" (not true), "if you don't like it get a Nvdia" (dumb), etc., etc.

 Helixhorned, on 09 November 2013 - 10:43 AM, said:

So, instead of complaining aimlessly, one would be better off providing experimental data that may be useful in identifying these cases of implementation-specified behavior, by say, doing comparisons of one particular scene on different systems.

shouldn't the people actually writing the software be doing this instead of using it as a crutch to not do anything to fix compatibility issues?

[Person of Color]

What Forge said. I've documented my issues with Polymer, including video, and only Terminx tried to help, IIRC no one else tried to really document it or anything.

Regardless, the consensus here is "filthy ATI user." Some of us want to own video cards that have better compute performance, more RAM, better memory bandwidth, and a lower price. Nvidia's current 600/700 series cards are total junk, they are designed for pure profit margin, and the only Nvidia card I've recommended in the past year or so is a 560 that was on clearance, and a 650 Ti that was on sale.

You guys need to get better Radeon support because the next gen consoles are all ATI, Mantle is coming out and will knock out some of DirectX's teeth, and games are in general going to be optimized for their architecture. Nvidia won't be the dominant player in a couple years. They dug their own grave by pissing off every company they've ever done business with.

I've owned far more Nvidia cards over the years, but they won't be getting my money any time soon. Especially with Mantle on the horizon.