mehere
New member
as can been seen in the pic xfire is on in all results
sorry if bettering your scores offends or you think I am cheating in some way
3DMARK Bench Thread
- Thread starter WYP
- Start date
as can been seen in the pic xfire is on in all results
sorry if bettering your scores offends or you think I am cheating in some way
agent_x007
New member
I was and still Am talking about Your Stock to Your OC scores only.
Stock : LINK
OC : LINK
I may be asking a lot, but please explain to me - HOW a ~45% CPU OC (from 3,5Ghz to 5GHz), with additional ~20% GPU OC (from ~1GHz to ~1,2GHz), can lead to a whooping 118% performance gain (ie. 2.18 times faster rig, with minimum gain of ~2.05x in EVERY aspect of the 3DMark11 benchmark - be it Graphics Score, Physics Score and/or Combined Score) - while using the exact same PC configuration all the time and for all tests [U say] ?
Basicly - I would like to know what sorcery U used to make this miracle happen
Last edited:
I wouldn't really say that locking the processor to it's base multiplier is truly 'stock' it should be left to Turbo to 3.9GHz as Intel intended. The purpose of ranking benchmark results by gains is to see who can do the best with what they have. Not who can gimp their stock score the best to make their gain look big!
agent_x007 I can see why your so confused but everything mehere posted is legit, there isn't any wizardry or CF tricks. With 4 cards the CPU clock speed will have a massive effect on the frame rate even in graphically intensive scenarios. His 'stock' scores are locked at the base multiplier of 3.5GHz while his CPU overclock is huge in the other tests. To be in line with the other people who have their results in the OP he should probably run the 'stock' tests with the CPU set to auto (3.5-3.9), letting the CPU boost to 3.9 is not an overclock as it is defined by Intel's spec.
Reading back through the last page what mehere posted to begin with was correct IMO, CPU was free to turbo and he chose to disable half the memory so he could clock it higher. That's perfectly reasonable he didn't make any hardware changes.
JR
agent_x007 I can see why your so confused but everything mehere posted is legit, there isn't any wizardry or CF tricks. With 4 cards the CPU clock speed will have a massive effect on the frame rate even in graphically intensive scenarios. His 'stock' scores are locked at the base multiplier of 3.5GHz while his CPU overclock is huge in the other tests. To be in line with the other people who have their results in the OP he should probably run the 'stock' tests with the CPU set to auto (3.5-3.9), letting the CPU boost to 3.9 is not an overclock as it is defined by Intel's spec.
Reading back through the last page what mehere posted to begin with was correct IMO, CPU was free to turbo and he chose to disable half the memory so he could clock it higher. That's perfectly reasonable he didn't make any hardware changes.
JR
Last edited:
agent_x007
New member
@JR23 U R telling me, a i7 3770 stock is bottlenecking four Radeon 7970s MORE, than a stock Celeron D 347 is bottlenecking a GTX 780 Ti ?
Graphic Score gain is in line with CPU Clock gain on my rig (Celeron D 347 and GTX 780 Ti) :
Stock : LINK
OC : LINK
Think for a second :
How much faster is i7 3770 over a Celeron D 347...
How much faster CFX of four 7970's CAN be than a one GTX 780 Ti card...
Question : Which configuration should benefit more from "extra MHz on CPU" ?
PS. Here's a bombshell for Ya, Fire Strike test :
Q6700 @ 3,69GHz with one GTX 780 Ti [STOCK] : LINK is FASTER, than a "stock" i7 3770 with "four" 7970's : LINK
So... How slow 4x 7970 can be "at stock" ?
I think CFX was on... but only two of the four cards were working during stock tests.
Graphic Score gain is in line with CPU Clock gain on my rig (Celeron D 347 and GTX 780 Ti) :
Stock : LINK
OC : LINK
Think for a second :
How much faster is i7 3770 over a Celeron D 347...
How much faster CFX of four 7970's CAN be than a one GTX 780 Ti card...
Question : Which configuration should benefit more from "extra MHz on CPU" ?
PS. Here's a bombshell for Ya, Fire Strike test :
Q6700 @ 3,69GHz with one GTX 780 Ti [STOCK] : LINK is FASTER, than a "stock" i7 3770 with "four" 7970's : LINK
So... How slow 4x 7970 can be "at stock" ?
I think CFX was on... but only two of the four cards were working during stock tests.
Last edited:
That's not truly his stock run though is it. Look how big the difference between 3.5 (8219) and 3.7 (12452) was, of course 5.2 plus massive GPU overclocks are going to have a profound effect.
Running 4 cards puts a ridiculous amount of strain on the CPU and it will throw up a lot of odd results for that reason. Just like in TTL's quad Matrix Platinum video some games performed worse with 4 cards than 2, it's not about processing power so much as shear speed and bandwidth. With Z77 as well it must be even worse, the MVE uses PLX chips etc unlike X79/X99. It's a little more unconventional than your typical single GPU and CPU dynamic which responds quite predictably to overclocking.
JR
Firestrike 1.1
JR23, 11.33% Gain, i5 4440 & R9 290X
Stock - http://www.3dmark.com/3dm/6187370?
OC - http://www.3dmark.com/fs/4279071
Running 4 cards puts a ridiculous amount of strain on the CPU and it will throw up a lot of odd results for that reason. Just like in TTL's quad Matrix Platinum video some games performed worse with 4 cards than 2, it's not about processing power so much as shear speed and bandwidth. With Z77 as well it must be even worse, the MVE uses PLX chips etc unlike X79/X99. It's a little more unconventional than your typical single GPU and CPU dynamic which responds quite predictably to overclocking.
JR
Firestrike 1.1
JR23, 11.33% Gain, i5 4440 & R9 290X
Stock - http://www.3dmark.com/3dm/6187370?
OC - http://www.3dmark.com/fs/4279071
Last edited:
mehere
New member
ran this again this morning everything on auto and with no voltage mods agent_x007
hopefully this will stop all the questions if not I'm done
stock P6900 http://www.3dmark.com/3dm11/9615275
Oc'ed P23827 http://www.3dmark.com/3dm11/9612331
gain = new - old
________ x 100 =245%
old
agent_x007
New member
I just want to say :
If that is indeed true - AMD drivers sucks REALLY Badly compared to NV.
To get just over 8,1k points on Graphics score in 3DMark11 (using sheer CPU bottlenecking), my stock GTX 780 Ti needed Penium D 840 @ 3,74GHz level of CPU performance : LINK, Screen : LINK (CPU-z : HT off = non-EE version)
And that Quad Fire setup can't even get to 8k Graphics score with a stock 3770k CPU ?
That's insane - what AMD is taking optimalisation for ?!
How much slower is Pentium D 840 @ 3,74GHz compared to i7 3770k ?
Simple answer : A LOT.
So... having A LOT more powerfull CPU, isn't enough to get all GPU's to work.
As for PCI-e lanes - Both stock and OC'ed results are affected (or should be affected) by this, and I really don't think it would show up more in non-OC result.
Last but not least, there's also this : LINK
As in : Best Hex Core/PCI-e/Quad Channel scaling U will ever see
If that is indeed true - AMD drivers sucks REALLY Badly compared to NV.
To get just over 8,1k points on Graphics score in 3DMark11 (using sheer CPU bottlenecking), my stock GTX 780 Ti needed Penium D 840 @ 3,74GHz level of CPU performance : LINK, Screen : LINK (CPU-z : HT off = non-EE version)
And that Quad Fire setup can't even get to 8k Graphics score with a stock 3770k CPU ?
That's insane - what AMD is taking optimalisation for ?!
How much slower is Pentium D 840 @ 3,74GHz compared to i7 3770k ?
Simple answer : A LOT.
So... having A LOT more powerfull CPU, isn't enough to get all GPU's to work.
As for PCI-e lanes - Both stock and OC'ed results are affected (or should be affected) by this, and I really don't think it would show up more in non-OC result.
Last but not least, there's also this : LINK
As in : Best Hex Core/PCI-e/Quad Channel scaling U will ever see
Last edited:
mehere
New member
I just want to say :
If that is indeed true - AMD drivers sucks REALLY Badly compared to NV.
To get just over 8,1k points on Graphics score in 3DMark11 (using sheer CPU bottlenecking), my stock GTX 780 Ti needed Penium D 840 @ 3,74GHz level of CPU performance : LINK, Screen : LINK
And that Quad Fire setup can't even get to 8k Graphics score with a stock 3770k CPU ?
That's insane - what AMD is taking optimalisation for ?!
How much slower is Pentium D 840 @ 3,74GHz compared to i7 3770k ?
Simple answer : A LOT.
So... having A LOT more powerfull CPU, isn't enough to get all GPU's to work.
As for PCI-e lanes - Both stock and OC'ed results are affected (or should be affected) by this, and I really don't think it would show up more in non-OC result.
Last but not least, there's also this : LINK
As in : Best Hex Core/PCI-e/Quad Channel scaling U will ever see
I'm now sick of arguing about this with you, I have posted several times in an effort to prove to you the results are good honest ones and you won't accept it
the results, the links and screen shots are all there
and as you said your not the OP, its not your call
accepted or not by the OP I done here
all I'm seeing now is you getting butt-hurt over some better scores than yours
and if you have to knock people to make yourself look better thats your life, you live it
if my results are good enough for Hwbot they are good enough for me http://hwbot.org/user/mehere1963/
Last edited:
agent_x007
New member
It's not "rearside" hurt.
That Quad Fire isn't working like it should.
6900pkt for Quad Fire in overall 3DMark11 P-Score "is OK" just because GPU-z and FutureInfo say CFX is working ?
Please... a Single 7970 can do 7k+ result in 3DMark11 P-Score test.
And if GPU scaling (ie. SLI/CFX) should work anywhere - it's in 3DMarks (that's the whole point of Quad Fire - tbh).
@joey I don't think AMD cards can do a "Tri-Fire" configuration.
Try using just two cards instead of three.
I'm done as well.
That Quad Fire isn't working like it should.
6900pkt for Quad Fire in overall 3DMark11 P-Score "is OK" just because GPU-z and FutureInfo say CFX is working ?
Please... a Single 7970 can do 7k+ result in 3DMark11 P-Score test.
And if GPU scaling (ie. SLI/CFX) should work anywhere - it's in 3DMarks (that's the whole point of Quad Fire - tbh).
@joey I don't think AMD cards can do a "Tri-Fire" configuration.
Try using just two cards instead of three.
I'm done as well.
Last edited:
I was asked to look at this because someone thought MeHeres results were low - but reading the thread its as if they are too high?
Anyways - Im totally lost - if you want to check any results look at the quad crossfire 7970 I did but remember the base system is different.
In regards to the discussion here - that is very welcomed - lets not make this a public trial. Being inquisitive is fine but dont cross the line boys. The same goes for replying to questions as to making them please.
Anyways - Im totally lost - if you want to check any results look at the quad crossfire 7970 I did but remember the base system is different.
In regards to the discussion here - that is very welcomed - lets not make this a public trial. Being inquisitive is fine but dont cross the line boys. The same goes for replying to questions as to making them please.
mehere
New member
Edit: New score.
System Specs:
- Intel Core i5 4670k @ 4602MHz
- XFX 7950 Boost crossfire (2) @ 1200/1610
Benchmark : Firestrike
Score (Stock): http://www.3dmark.com/3dm/7856876
Score (overclocked): http://www.3dmark.com/3dm/7868100?
http://www.3dmark.com/fs/5477305 Percentage gain (%): 35.06%
Not beating agent x007 with his 75% gain but I'mm happy with the result
Also I left boost on the CPU so it boosted from 3.4 to 3.8, was I suppose to do that? ^_^
As for mehere's score, he probably had core 1 clocked high and the rest low, again with GPU's, probably had GPU one clocked high, rest below stock.
There is NO way a stock 3770k and 4xcrossfire 7970 ONLY scores 8200 in Firestrike, stock i5 4670k and 7950 crossfire (2) scores 9644, stop playing us like we are 6 years old. It's pathetic.
System Specs:
- Intel Core i5 4670k @ 4602MHz
- XFX 7950 Boost crossfire (2) @ 1200/1610
Benchmark : Firestrike
Score (Stock): http://www.3dmark.com/3dm/7856876
Score (overclocked): http://www.3dmark.com/3dm/7868100?
http://www.3dmark.com/fs/5477305 Percentage gain (%): 35.06%
Not beating agent x007 with his 75% gain but I'mm happy with the result
Also I left boost on the CPU so it boosted from 3.4 to 3.8, was I suppose to do that? ^_^
As for mehere's score, he probably had core 1 clocked high and the rest low, again with GPU's, probably had GPU one clocked high, rest below stock.
There is NO way a stock 3770k and 4xcrossfire 7970 ONLY scores 8200 in Firestrike, stock i5 4670k and 7950 crossfire (2) scores 9644, stop playing us like we are 6 years old. It's pathetic.
Last edited:
