Look forward to seeing it if you do.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
** KINGPIN HAS ENTERED THE WEBSITE!! **
- Thread starter Gibbo
- Start date
I don't buy that . Sure, it's good for LN2 but air and water are also listed on the Kingpins overclock guide/sheet.
Also, I don't buy that these cards "are binned" for LN2. Both of my regular 780Ti Classifieds do better than my Kingpins. If the Kingpins have lower core freq before artifacting, and at a higher voltage than my 780Ti's all on water, how is that better for LN2? Logic would dictate a great performing card on air and water would lead to a great LN2 card, not a poorer performer becoming a great LN2 card.
Anyways I don't LN2, so for gaming under high end water the Kingpins are a letdown.
Very sad to hear
There is nothing worse than getting stuff running only to find out that it does not live up to expectations.
What are you going to do, stick with the KPE's or change to another setup.
Soldato
- Joined
- 5 Sep 2011
- Posts
- 12,881
- Location
- Surrey
Do we perhaps think there is something going on with this pcb that is creating this compromise on the memory? Although if that were the case why bother paying out on Samsung chips.
People are automatically assuming there is a problem to fix. It could be that the Samsung chips were selected for their ability to provide stability rather than speed when the core is at LN2 clocks and voltages.
Good point - never thought of it that way
Well then why is EVGA investigating this and testing out various bios at the moment? That last LN2 run on HWBOT has the memory at 7400 MHz- less than the 7800 indicated on the EVGA KPE overclocking guide which should have been a conservative guide at best. No matter the voltage applied, people are unable to increase the memory past a certain point as I mentioned earlier.
Last edited:
Soldato
- Joined
- 5 Sep 2011
- Posts
- 12,881
- Location
- Surrey
+1 good point. Although obviously there's a bit of a misunderstanding as more importantly owners (vega included) are a little stumped by the memory clocks they're getting.
Ya, I don't think anyone that has a Kingpin has been getting even the OC guide stated memory freq's stable.
I also think Nvidia limitations are coming into play somehow. Black screen prevalent when you clock both mems and core. Mems on my Classy will do 2050 - 2100 air but not so with 1600+ core black screen memory problem. I am not convinced at all Nvidia drivers are not in some way a limiting factor to getting the best from even Kingpins cards.
If you notice for the first time ever, KPE card XOC bios has no boost. Normally Vince locks boost with K Boost (NVP manager) we all do. This also locks memory clocks in 3D. With Nvidia ti dribvers you cant use NVP manager on anything but there limited bios (1.21v). You get a blue screen on reboot. With no boost Bios KPE core is locked once the bench is entered but memory fluctuates between 2D and 3D in GT's etc causing instability.
On DC2 you can remove a resistor to lock memory in 3D all the time. Maybe this is why DC2 seems to be better clocking on mems. I am going to mod a bios for DC2 to lock boost also as KPE and see where this takes me with resistor removed.
Just a theory I have. I must test much more to be certain.
If you notice for the first time ever, KPE card XOC bios has no boost. Normally Vince locks boost with K Boost (NVP manager) we all do. This also locks memory clocks in 3D. With Nvidia ti dribvers you cant use NVP manager on anything but there limited bios (1.21v). You get a blue screen on reboot. With no boost Bios KPE core is locked once the bench is entered but memory fluctuates between 2D and 3D in GT's etc causing instability.
On DC2 you can remove a resistor to lock memory in 3D all the time. Maybe this is why DC2 seems to be better clocking on mems. I am going to mod a bios for DC2 to lock boost also as KPE and see where this takes me with resistor removed.
Just a theory I have. I must test much more to be certain.
Last edited:
That's a pretty good theory, and makes sense as well. If it does turn out that the Nvidia drivers are the ultimate bottleneck to overclocking, then I don't know if I can ever bring myself to buy an Nvidia card.
Latest I heard from EVGA is that they are presently testing ~20 different memory timings to see if that helps with the overclocking. I don't think they anticipated this and people in the US (including me) are still undecided whether or not to keep the cards at the moment. I might end up binning both my cards, keep one and sell the other at this point.
Latest I heard from EVGA is that they are presently testing ~20 different memory timings to see if that helps with the overclocking. I don't think they anticipated this and people in the US (including me) are still undecided whether or not to keep the cards at the moment. I might end up binning both my cards, keep one and sell the other at this point.
Ok, just did some exhaustive testing with a Kingpin vs regular Ti Classified battle. All the power settings are the same between the cards. I did these tests in 3x Portrait mode, as that is what I use for gaming. The figures are the best the cards could do artifact free, and memory crash free. These are the average of the benchmarks:
My best Kingpin: ------- 1424 Core / 7600 Mem / 116.3 FPS
Ti Classified Stock Mem: 1424 Core / 7000 Mem / 110.2 FPS
Ti Classified Same Mem: 1424 Core / 7600 Mem / 114.9 FPS
Ti Classified Max Mem: 1424 Core / 8400 Mem / 119.7
So, a few things that came out of this. My best Kingpin does the same as my best Ti Classified core. My other Kingpin does less core. Same core and same Mem freq between them, the Kingpin ended up being 1.2% faster. So the Samsung memory does have slightly tighter timings. Now bump up the Classified's Mem to it's max, and I get a 3% gain in performance over the best my Kingpin can do.
So it comes down to my Ti Classified being the fastest card I have. I will hold onto the Kingpin to see if EVGA can work on the memory situation, it's whats really holding this card back.
Oh, I also found out that 600 KHz PWM on the Kingpin actually raises the under load voltage higher than the same setting on the Ti Classified. As an example, 1.35v on EVBOT gives DMM 1.45v on the Kingpin, and 1.35v on EVBOT gives 1.42v on the Ti Classified.
My best Kingpin: ------- 1424 Core / 7600 Mem / 116.3 FPS
Ti Classified Stock Mem: 1424 Core / 7000 Mem / 110.2 FPS
Ti Classified Same Mem: 1424 Core / 7600 Mem / 114.9 FPS
Ti Classified Max Mem: 1424 Core / 8400 Mem / 119.7
So, a few things that came out of this. My best Kingpin does the same as my best Ti Classified core. My other Kingpin does less core. Same core and same Mem freq between them, the Kingpin ended up being 1.2% faster. So the Samsung memory does have slightly tighter timings. Now bump up the Classified's Mem to it's max, and I get a 3% gain in performance over the best my Kingpin can do.
So it comes down to my Ti Classified being the fastest card I have. I will hold onto the Kingpin to see if EVGA can work on the memory situation, it's whats really holding this card back.
Oh, I also found out that 600 KHz PWM on the Kingpin actually raises the under load voltage higher than the same setting on the Ti Classified. As an example, 1.35v on EVBOT gives DMM 1.45v on the Kingpin, and 1.35v on EVBOT gives 1.42v on the Ti Classified.
Soldato
- Joined
- 5 Sep 2011
- Posts
- 12,881
- Location
- Surrey
Well you'd know better than most lets be honest! I think we need to see some LN2 results too just to see if the Sammy chips do any better in sub zero. Although I'm not sure really if that puts bed to the argument or not for those on water.
If I can game on my cards at anywhere near 1.4GHZ I will be in awe. I don't care much for RAM clocks, stability is more important for me.
I think expectations need to be realistic, if the cards can reliably do 1.4GHZ without any artifacts under water and under warranty then that makes them pretty epic in my book.
Peeps that already had classifieds which do these kind of clocks were being unrealistic if they expected 1.5 or 1.6GHZ out of these cards without sub-zero cooling.
I think expectations need to be realistic, if the cards can reliably do 1.4GHZ without any artifacts under water and under warranty then that makes them pretty epic in my book.
Peeps that already had classifieds which do these kind of clocks were being unrealistic if they expected 1.5 or 1.6GHZ out of these cards without sub-zero cooling.