both.

ring/cache voltage. he mistook it for vccin.

but do u guys even know what that voltage is for??..
pretty sure some massive google will happen after i post this

download 3dmark11.. errr ahem version if u can
go to advance



mistake untick demo... forgot to do that.. just on graphic test 1

set the reso u want
DO not able MSAA it will reduce the stress..

have fun

and games tomb raider is very stressfull etc..

This post has been edited by cstkl1: May 23 2014, 12:35 PM

u should see the stress on ure vrm powerlimit rather than just gpu.
last time tested this on ati was 7970ghz sapphire trixx.

on nvidia.. lol it will hit more than at 300watts power limit at 1.05v only
so when u keep looping it.. if u can see ure gpu voltage and vrm temps and powerlimit... u will see any flux.



This post has been edited by cstkl1: May 23 2014, 09:24 PM

second test is tomb raider max with tressFX. ure case with crossfire.. u should one card at a time.

heck if u can even do downsampling at 4k to push the limit on each card dvi bandwidth.

This post has been edited by cstkl1: May 23 2014, 09:27 PM

that was few min run.. it already spiked to 95% at only 1.05v dude. on a custom bios. Default bios is 250 watts at 100%. this is 300watts at 100%.
hence y i dont run unless benching on higher.. and after 110% need water to do 1300mhz etc but at 140%.. thats 420watts....

.

CPU Dependent Benchmark that is.

4770k. Each multi goes up u get 700-1k score on overall. GPU even raise it on graphic score 10k more.. not much on overall in comparison.

http://www.3dmark.com/sd/2133408

so lol a 4930k with single card is gonna whack any 4770k with dual cards easy.


This post has been edited by cstkl1: Jun 12 2014, 04:21 AM

really dont bother about that gpu score. it contributes very little to the overall.
a joker with 4730k 5ghz etc can own this benchmark with a single card.

yeah i really didnt look at it.
cause nvidia driver 337.88 .. graphic test 1 is totally blank with only fps counter.
the rest works fine.

any FM guy said anybody above 20k should use firestrike
and if ure 20k on firestrike u should use extreme.

did u notice now 3dmark is like 3gb...

Somewhere in the middle he admitted it doesnt hurt AMD now but they assume it WILL one day.

The only time a actual game and example was given was crysis 2 on tessellation. Afaik tessellation is a DX11 thing. So why they dont fix it on their cards.

Basically he is argueing on one point only
1. According the Game Developer email conversation they had.. they were not allowed access
Based on this they came up with nvidia is crippling games with their tech and amd suffers worse from this cripple.

2.I wonder why would a game developer even endorse this?? End of the Day they are the one will get the bad review etc like how ubisoft is getting hammered basically every hour for the last few weeks on every media/social avenue available.
Watchdog only used HBAO from gameworks and it doesnt even affect performance at all for both camps.

3.Mantle is still a closed API that is heavily guarded. No sharing there. FreeSync Requires better panels and response time and higher cost. Gsync until today theres no proper reviews on it so seriously why that extra buffering etc is needed.. is really unknown.
Of course Intel will be against mantle as this will take a chunk of their market as " Gaming CPU".

4.Only good thing i hear from this is they are sharing their tech with Microsoft which will make DX12 realization sooner.

5.Btw if i am a stockholder of AMD shares... I would be lambasting them for using my dough to do FREE things. Although US tax policies does allow claims for R&D... but seriously all i can see them doing this is because
of CPU's .. bottleneck issue.. So we all can buy their cpu's now since its cheaper. Finally they can compete with Intel more on Core Basis that threaded performance

And i think he was talking about DDR4.

really would like know
empire23
comment on this video.

This post has been edited by cstkl1: Jun 20 2014, 05:51 PM

This time he was clearer.

Have to say it. Nvidia are scums.

are u using hdmi or dvi

try different hdmi cables

but if possible try a dvi monitor or borrow one etc.

to rule out its the VRAm or its controller issue.

i know but theres three 4 parts in the GC, GPU, VRAM, VRM and Display controller.
So need to check which is causing it.

from what ure saying sounds its a VRM issue.

http://videocardz.com/50947/amds-radeon-r9...might-full-chip

8pack said it.

and now confirmed.
http://videocardz.com/50956/confirmed-amd-...-hawaii-xtx-gpu

3k core..

so hawai xt was not a full chip

if this pair this with 7ghz mem and 8gb
i am selling my blacks.


This post has been edited by cstkl1: Jun 25 2014, 05:50 PM

if i know nvidia...

780ti Ghz version. As u all know already the new 110 fab is cooler, and uses less voltage for 1.2ghz (1.05-1.1v) ( The 1254mhz barrier at is the mem controller paired with the 7ghz mem ( downclock and it will scale even higher) which is why KPe uses samsung i think.. ( seems to have less impact with the mem controller even with tighter timings.. i guess its some like PSC vs elpida hyper kinda of thing.. psc on X58 requires less IMC voltages to drive at the same CL and RTL's)

so interesting time this time. Both companies are top of their game. sooner or later it has to come down to price wars.

This post has been edited by cstkl1: Jun 25 2014, 06:46 PM

show a picture of the loop.
btw what pump??
ek doesnt mind if u do a reverse flow on the gpu .. aka vrm to gpu on in and out but affects temps on multi card config.

so ure in and out does matter on some blocks.
edit:
well just checked the aqua block .. looks like doesnt matter.

so the issue here is ure loop flow config.

knowing a lot of ppl with DDC.. they will do

DDC to Rad .. question is which rad u did it with.

cause u have the best block+backplate cooling solution for the 290x.

btw bro WC .. stuff like this common. Heck i pernah kena worse stuff like pump oring, mobo oring, gunk inside, bleed issue...

really need to see a pic. if u can send to me via pm.


This post has been edited by cstkl1: Aug 5 2014, 07:03 PM

aqua blocks are easy to install.

They use tim, built in standoffs.

The issue here is either there is a qc issue with his block or he has a rogue gpuc pcb that is different or issue with flow.

I suspect the last or hmm his backplate is the active one... could be pushing the mainblock out a bit hence less contact.

Sentinel92
Try not using the backplate etc.

ok i pretty sure its a flow rate issue

Blocks love high flowrate, rads love low flowrate.
Flowrate is inverse to Head pressure.
Using angle adapters, fittings, blocks etc will reduce flowrate and increase headpressure
and headpressure makes it go further enough to complete the flow back to the pump..

So with ure DDC.
Its has low flowrate and high headpressure

Look at ure loop u did Pump to CPU to a 420 rad and GPU and 240.

Having that 420 there kills flowrate which ure gpu needs. And ure 420 is getting higher flowrate than the 240. so efficiency in cooling ure 240 is doing more work.

So u should try

Pump to CPU to GPU to rad 420 and 240 and then to res. Ure ddc has already limited flowrate. and your cpu block afaik is restrictive.

Again theory. Only way to test is try.


This post has been edited by cstkl1: Aug 5 2014, 09:20 PM

hence y always plan with a fillport and drainport.