Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [red quasar] First benchmarks of Haswell non engineering samples.
New Posts  All Forums:Forum Nav:

[red quasar] First benchmarks of Haswell non engineering samples. - Page 19

post #181 of 276
Even at 4.5 it is not a stable oc.
Quote:
超4.5G bios 电压1.25V 跑Prime95 蓝~ 1.26 继续蓝 1.28蓝的更快~ 1.30直接蓝

Translate to:
The benches were run with 4.5G oc. Prime 95 blue screen @1.25v, still crash @1.26 and even crash faster @1.28, and it can't even run at 1.30v.

Very disappointing indeed.
post #182 of 276

There are Ivy chips that take 1.35V for 4.5 (and plenty that take 1.3V) and Haswell is on the same process, so just because one chip couldn't do 4.5@1.4 doesn't mean it's going to be a disaster.  I am hoping for better though.

 

 

Quote:
Originally Posted by tpi2007 View Post

Yeah sorry, it's a bit late here, but anyway, isn't 10% over 4 year more like 50% improvement and not double ?

 

Yeah, 7 years to double at 10% a year.


Edited by Forceman - 5/9/13 at 9:33pm
post #183 of 276
Quote:
Originally Posted by tpi2007 View Post

Yeah sorry, it's a bit late here, but anyway, isn't 10% over 4 year more like 50% improvement and not double ?

10% of 100 is 110 10% of 110 is 121 so on and so forth, by 5 years, its more then doubling.


Quote:
Bulldozer was a screw-up when you look at AMD's own offerings, you don't even need to look at Intel's. When the FX-4100 is basically not an upgrade for any Phenom II X4 owner and yet uses the same or more power, the FX-6100 isn't an upgrade to X6 1090T / 1100T owners and not even the FX-8150 is much of an upgrade to X6 1090T / 1100T owners more like a sidegrade, you have to question their execution. You may talk about the future and how it lays the foundation for better things to come, blah, blah, blah, but that doesn't conflict with the fact that their first try was underwhelming.
you miss the point, bulldozer wasn't overwhelming, it just didn't work will with current software, it smashs the phenomii in more then a few areas to.

Quote:
It's not confidence in Intel's infallibility in general, which is what I take you're trying to take from my post, I'm only talking about Haswell, and in this case it's the mere fact that Haswell is based on IB, this isn't a whole new architecture like P3 -> P4 or like Phenom II -> Bulldozer, where both companies are taking big chances, release a product before it's truly ready (same thing happened with the first Pentium 4 Willamette) and then spend the next few generations polishing it, so claiming that Haswell will perform worse than IB would take some serious thing going wrong, which is highly unlikely. Sure, they might have relaxed a few timings in favour of more power efficiency, but that is more believable in laptops, there is no sense in doing that on a desktop CPU where the additional power savings from doing that are not worth the potential tarnish to the brand image.

I don't know if you understand this, but the days of big uarch changes netting additional ipc, are comming to a close.
post #184 of 276
Ok look a few pages back u will see the first chip they used its beat all to heck the IHS looks like it got run over. I posted up the pics from the site then they tested a different chip with better results.
Primo
(15 items)
 
  
CPUMotherboardGraphicsRAM
I7-7700k Asus Maximus Hero ix Evga GTX 1080 ACX 3.0 G-Skill F-43200C15D-16GTZKW 
Hard DriveHard DriveCoolingCooling
Samsung 850 Evo Corsair Neutron  Ek Supremacy CSQ clean OCOOL 360mm radiator 
CoolingCoolingOSKeyboard
OCOOL 360mm radiator Ek -FC1080 GTX  Windows 10 CM Storm 
PowerCaseMouse
Evga 750 watt Phanteks Enthoo Primo Roccat Kone Optical 
  hide details  
Reply
Primo
(15 items)
 
  
CPUMotherboardGraphicsRAM
I7-7700k Asus Maximus Hero ix Evga GTX 1080 ACX 3.0 G-Skill F-43200C15D-16GTZKW 
Hard DriveHard DriveCoolingCooling
Samsung 850 Evo Corsair Neutron  Ek Supremacy CSQ clean OCOOL 360mm radiator 
CoolingCoolingOSKeyboard
OCOOL 360mm radiator Ek -FC1080 GTX  Windows 10 CM Storm 
PowerCaseMouse
Evga 750 watt Phanteks Enthoo Primo Roccat Kone Optical 
  hide details  
Reply
post #185 of 276
Quote:
Originally Posted by SCollins View Post

10% of 100 is 110 10% of 110 is 121 so on and so forth, by 5 years, its more then doubling.

You might want to do that math again.

 

100

110

121

133

146 (year 4)

161

177

195 (year 7)

post #186 of 276
Its 2 chip, not one. The first chip they got have similar, if not exactly the same, problem, iirc.
Well, both were same C0, but from different batch, so my hopes are not high for this one.
post #187 of 276
Quote:
Originally Posted by OutlawII View Post

Ok look a few pages back u will see the first chip they used its beat all to heck the IHS looks like it got run over. I posted up the pics from the site then they tested a different chip with better results.

Yes, since the first chip looks like being abused, they got a new boxed chip, instead of a tray one, to do the test and to proof if it was the problem with the IHS.

They have overheat problem with the first chip and now the new chip is not doing much better, but they did not post temperatures so it could be the same problem.

The results are no better anyway.
post #188 of 276
Quote:
Originally Posted by Forceman View Post

You might want to do that math again.

100
110
121
133
146 (year 4)
161
177
195 (year 7)

I guess that not everyone learned about the "Rule of 70" (or 72) in ECON 101...
Julia
(11 items)
 
Centurion (Dead)
(11 items)
 
 
CPUMotherboardGraphicsRAM
AMD Athlon 64 3700+ EPoX EP-9NPA+ Ultra EVGA GeForce 6800 GS (256-P2-N389-AX) CORSAIR VS1GB400C3 
Hard DriveOptical DriveOSMonitor
Western Digital Caviar SE WD1600JS NEC ND-3520A Windows XP Dell M780 
PowerCaseOther
FSP Group SAGA+ 450R Cooler Master Centurion 5 w/ Window (CAC-T05-WW) Samsung Floppy Drive (SFD321B/LBL1) 
  hide details  
Reply
Julia
(11 items)
 
Centurion (Dead)
(11 items)
 
 
CPUMotherboardGraphicsRAM
AMD Athlon 64 3700+ EPoX EP-9NPA+ Ultra EVGA GeForce 6800 GS (256-P2-N389-AX) CORSAIR VS1GB400C3 
Hard DriveOptical DriveOSMonitor
Western Digital Caviar SE WD1600JS NEC ND-3520A Windows XP Dell M780 
PowerCaseOther
FSP Group SAGA+ 450R Cooler Master Centurion 5 w/ Window (CAC-T05-WW) Samsung Floppy Drive (SFD321B/LBL1) 
  hide details  
Reply
post #189 of 276
Quote:
Originally Posted by OutlawII View Post

Ok look a few pages back u will see the first chip they used its beat all to heck the IHS looks like it got run over. I posted up the pics from the site then they tested a different chip with better results.
That was my first thought as well. Maybe someone used a hammer to get to the TIM (delidding done wrong) to make pictures? Then they put it back on and oops. The CPU did no longer perform as well as it used to do.
post #190 of 276
Quote:
Originally Posted by skyn3t View Post

dang i think i will wait too. I don't need another hotter cpu one after another. what sup intel.?

 

~flies away with 3930k ~ : )

 
≡ Snowbaru ≡
(17 items)
 
 
CPUMotherboardGraphicsGraphics
Core i7 3930K 4.9Ghz Asus Sabertooth X79 EVGA GTX TITAN 1250Mhz KPE/Hybrid BIOS (SOLD) NVIDIA Geforce GTX 1080 Ti Founder Edition 
RAMHard DriveHard DriveHard Drive
32GB Samsung 30nm 2400Mhz OCZ Vertex 4 Samsung 850 EVO 13000GB HDDs 
CoolingOSMonitorKeyboard
EKWB Custom Watercooling, XSPC, Bitpower, Primo... Windows 10 Pro x64 (1607) ASUS PG278Q G-SYNC (2560x1440) Logitech G910 Onion Spark 
PowerCaseMouseMouse Pad
EVGA SuperNOVA 850G2 Gold Fractal Design Define S Logitech G502 Logitech G440 
AudioOtherOtherOther
Logitech G933 Xbox One Controller Asus RT-AC68U Hitron CDA3 
CPUCPUMotherboardGraphics
5HP 5HP 5HP 5HP 
RAMHard DriveOptical DriveCooling
5HP 5HP 5HP 5HP 
OSMonitorKeyboardPower
5HP 5HP 5HP 5HP 
CaseMouseMouse PadAudio
5HP 5HP 5HP 5HP 
Other
5HP 
  hide details  
Reply
 
≡ Snowbaru ≡
(17 items)
 
 
CPUMotherboardGraphicsGraphics
Core i7 3930K 4.9Ghz Asus Sabertooth X79 EVGA GTX TITAN 1250Mhz KPE/Hybrid BIOS (SOLD) NVIDIA Geforce GTX 1080 Ti Founder Edition 
RAMHard DriveHard DriveHard Drive
32GB Samsung 30nm 2400Mhz OCZ Vertex 4 Samsung 850 EVO 13000GB HDDs 
CoolingOSMonitorKeyboard
EKWB Custom Watercooling, XSPC, Bitpower, Primo... Windows 10 Pro x64 (1607) ASUS PG278Q G-SYNC (2560x1440) Logitech G910 Onion Spark 
PowerCaseMouseMouse Pad
EVGA SuperNOVA 850G2 Gold Fractal Design Define S Logitech G502 Logitech G440 
AudioOtherOtherOther
Logitech G933 Xbox One Controller Asus RT-AC68U Hitron CDA3 
CPUCPUMotherboardGraphics
5HP 5HP 5HP 5HP 
RAMHard DriveOptical DriveCooling
5HP 5HP 5HP 5HP 
OSMonitorKeyboardPower
5HP 5HP 5HP 5HP 
CaseMouseMouse PadAudio
5HP 5HP 5HP 5HP 
Other
5HP 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [red quasar] First benchmarks of Haswell non engineering samples.