I glad I bought R9 290X before playing GTA V. The card is absolute BEAST. Even better I been running the card at measly 1000MHz core and 1000MHz memory (VDDC offset is set at -93mV) yet the performance doesn't drop one bit! It eats measly 148W VDDC In and 125W VDDC out from GPU-Z. My system is as quiet as it was idle at this clock It still scored 127xx in 3DMark11 at this clock.

You need one as a proof? Im too busy playing GTA V

No, Im using stock BIOS. Your BIOS clocks are a lot higher but it also adds +25mV which I don't like. Here is the screenshot running 3DMark11 with 1GHz speed

Mine is quite low though, 69.6%. What's yours?

GPU-Z reading is inaccurate. Someone got over 100% ASIC Quality which is impossible

This was my pride and joy back then, X1800GTO. Costs me RM1000 back then. I just love reference designed PCB, there is some sort of elegance into it. Back then I was foolish, I overclocked it way too much that it reached over 100C for quite some time and killed it

Earlier batches are possible but they are very rare nowadays. Powercolor, XFX, Sapphire very early cards are able to unlock, you can check out this thread for more info on unlocking.

That looks like artifacts usually occur because of bad VRAM. You don't encounter this problem on other game?

True words has been spoken.

GPU temperature is still fine by your screenshot. Running benchmark is one thing, you need to play games because some games isn't stable even though benchmark is. But whatever you do don't ever EVER run Furmark.

It sad to say that you missed the good CPU overclocking days. Judging from current trend, overclockable CPU will be less and less. I hope AMD somehow will release entire range overclocking like on AM3+ FX CPU on their upcoming Zen arch. Multiplier overclocking is boring too, you don't play with memory divider and play with timings when changing the bus speed anymore

Back then overclocking CPU give massive performance improvement and massive clockspeed increase. I once overclocked socket 478 Northwood core Pentium 4 2GHz all the way to 3GHz on stock cooler and cap ayam PSU running The real gem is the Albatron motherboard I'm using. PX845E and PX845G. Great motherboard. Hey I even got the screenshot of it With my 6800LE unlocked all 16 pipeline! Those are the days



The second gem is when I transitioned to socket 754 buying the cheapest Palermo core Sempron 2500+ with clockspeed of mere 1.5GHz stock. Using DFI nF4X Infinity motherboard I once again reach the amazing 1GHz overclock to 2.5GHz again on stock cooler! At stock 1.5GHz speed in benchmark this is faster than my Pentium 4 at 3GHz in ANY benchmark, you can imagine how fast this is at 2.5GHz. I have the screenshot of it too



My current i5 4670K also reaches the 1GHz overclock target, but I can't use stock cooler anymore

Back then I'm overclock because I can't afford higher end CPU and GPU, overclocking closes up the gap and makes things more enjoyable. I also play considerably longer back then because I feel an enjoyment after all the tweaks been done. Now that I'm able to buy good stuff makes me play less actually, it's either the games is not enjoyable anymore or it's just I'm getting old

Since you guys talk about driving game, I do play driving games back then, I play a lot of CMR 2004 with old Creative Avant Force NX wheels. Since that broken I never play driving games anymore

GLide looks absolutely incredible on Unreal as compared to Direct3D. I still have 3DFX Voodoo Banshee but I don't have PC with AGP 2X slots, only 4X slots so I can't use it anymore

It puts crazy amounts of load, which is no current games or other benchmark puts on the GPU. It's a stupid program runs by idiots. Just don't do it.

GPU overclocking is as easy as moving the slider on the program. CPU overclocking is more than that. This is how I learn how computer works; what FSB do, what multiplier is, what is memory timing and divider etc etc etc.

Power consumption back then is not so much as today hardware thus cap ayam PSU can run an overclocked system.

Fun? Hell yeah, it's the best times where you tweak all day and play all night. Today's overclocking is a little bit different but it's not as much fun because most cheap CPU are locked and you need to pay extra for the unlock ones, plus overclockable motherboard chipset (yuck). The purpose of overclocking is buy cheap hardware and overclocked it to match a much faster hardware, at least that is on my book.

Meh we been stuck at around 4GHz for SO LONG now it's not even funny. You won't even see 10GHz, let alone 1000GHz

Google 'Intel Tejas' then you know why I say you won't even see 10GHz. Intel and AMD abandoned the MHz race (somehow AMD revert back to this speed race with Bulldozer arch) but they will move back to IPC and CMT/SMT on next Zen arch. Means they abandoned pure speed for more efficient multi core CPU at reasonable clockspeed that pumps more IPC. Intel once predicts reaching 10GHz by 2011. It's been 2015 and the fastest Intel CPU is 4.4GHz and that is it's boost state.

Weird 15.4 driver dont give any problems to me. Have you checked your temperature? GPU wont throttle without reason

Older games works but if you really read whats the problem with new games specifically Project Cars you know it wont be that easy. Updating PhysX to latest version makes the AMD + nvidia PhysX combo useless. It worked on older games but not on newer games. Nvidia locks it at physx level, nice isnt it?

I think you confused, GLide is another API created by 3DFX which does not interfere with other renderer or makes other slower. What physx done is something else, it runs well on nvidia cards and make it suck on other hardware. Some games you can disable it but for the hot topic Project Cars you cant. And thats annoying

It's true to some degree but back then Direct3D and OpenGL isn't as good as GLide because of large CPU overhead, all I can say it's like AMD Mantle of today, it is a low level API that took advantage of hardware much better than DirectX or OpenGL. And because of it, DirectX and OpenGL become more mature, as in today where AMD Mantle 'forced' the two big API to go low level to get lower CPU overhead and increasing performance on their next iteration.

I don't mind if the games like Borderlands series which I could set the PhysX setting to my liking but on Project Cars PhysX used in everything (rain, car balance on corners, tire grip on road etc.) so it can't be disabled. It's like nvida Gameworks just force game devs to use all of it's hardware on their sponsored games. On the contrary, AMD TressFX uses DirectCompute which is available for nvidia as well but being nvidia, some generations (like Kepler vs Fermi) the compute performance actually goes DOWN instead of up. AMD GCN are designed for compute in mind.

I don't see any DVI ports, maybe purely DP for six display perhaps?

Anyhow I'm still skeptical about AMD is going with HBM going all high-end all of the sudden. Thing is I been testing my 290X for it's memory clock scaling. I didn't notice much improvement even if I downclocked the memory by quite a significant margin (1GHz (4GHz effective) still only loses few fps). Unless AMD feeding the new Fiji with 96 ROP or 128 ROP I don't think the core is starving on bandwidth. They could always test the new HBM on lesser GPU like they did on HD 4770 where they first test GDDR5 memory.

I don't know what's the fuss over tessellation, you can configure tessellation detail in CCC. IMO the visual effect of it isn't subtle and you won't miss it.

The number isn't out yet so stop with the 'If' thing. Just wait. It's not going to be long.

Exactly. It won't be just clockspeed, it is a die harvested GPU. I got the feeling the number of CU and subsequent ROP/TMU is cut down significantly to achieve it. I'd keep my 290X. I could always downclock it to achieve lower power consumption, easy peasy. Still, I reserve my judgement until real numbers comes out