Overclock.net › Forums › General Hardware › General Processor Discussions › Why are we stuck at 3-4 ghz ???
New Posts  All Forums:Forum Nav:

Why are we stuck at 3-4 ghz ??? - Page 2

post #11 of 18
Quote:
Originally Posted by Snowmen;15331676 
^ This.

IBM made CPUs out of graphene that were running at 100GHz.

Graphene cpu can't turn off.. according this article. http://www.tgdaily.com/hardware-features/53714-ibm-graphene-wont-replace-cpu-silicon

We're definitely stuck
    
CPUMotherboardGraphicsRAM
Xeon e5520 DX58SO Radeon 7850 3x2gb OCZ Gold 
Hard DriveCoolingOSMonitor
1tb Hitachi Deskstar 7200rpm CoolerMaster N520 Windows 7 Home Premium x64 Westinghouse 22" 1680x1050 
PowerCaseMouseAudio
Earthwatts EA-500 500w Thermaltake WingRS Steelseries Sensei RAW HT Omega Striker 7.1 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Xeon e5520 DX58SO Radeon 7850 3x2gb OCZ Gold 
Hard DriveCoolingOSMonitor
1tb Hitachi Deskstar 7200rpm CoolerMaster N520 Windows 7 Home Premium x64 Westinghouse 22" 1680x1050 
PowerCaseMouseAudio
Earthwatts EA-500 500w Thermaltake WingRS Steelseries Sensei RAW HT Omega Striker 7.1 
  hide details  
Reply
post #12 of 18
Assuming electricity is going the speed of light, (it is actually somewhat slower)
a signal can only travel a couple inches @ 3ghz, and the signal may have to make more than a one pass inside the chip.

That hardly limits them to 3ghz, but a speed like 100ghz wouldn't seem possible.
Edited by justinjja - 10/16/11 at 11:59am
Lappy
(13 items)
 
  
CPUMotherboardGraphicsHard Drive
I5 2520M Dell Intel HD3000 Samsung 830 128GB SSD 
OSMonitor
Windows 7 Ultimate LED 
  hide details  
Reply
Lappy
(13 items)
 
  
CPUMotherboardGraphicsHard Drive
I5 2520M Dell Intel HD3000 Samsung 830 128GB SSD 
OSMonitor
Windows 7 Ultimate LED 
  hide details  
Reply
post #13 of 18
Quote:
Originally Posted by Snowmen View Post
^ This.

IBM made CPUs out of graphene that were running at 100GHz.
IIRC it wasn't a CPU but 1 or a couple transistors
ShadowForge
(12 items)
 
Defiant
(14 items)
 
CarbonCat
(13 items)
 
CPUMotherboardGraphicsRAM
Phenom II x6 1405T (unlocked Athlon II X4 640T) ASUS M5A99X EVO AM3+ Asus ENGTX470/2DI/1280MD5/V2 16 GB (4x4GB) G.Skill DDR3 1600 CAS9 1.35v 
Hard DriveHard DriveOptical DriveCooling
WD Caviar Blue 250 2.5" Laptop Drive WD Caviar Black 1TB 3.5" Pioneer BDR-203 BluRay Burner Corsair H60 push 
OSOSMonitorPower
Windows 7 Pro x64 Ubuntu 11.10 Samsung 40" 60hz  ANTEC NEO ECO 520W 
CaseMouse
NZXT Gamma Microsoft Bluetooth Notebook Mouse 5000 
CPUMotherboardGraphicsRAM
i7 - 2600k [5.0 1.42v] ASUS P8Z68 Deluxe Sapphire HD6950 2gb Dirt 3 Edition 8GB G.Skill DDR3 2133 CAS11 
Hard DriveOSMonitorKeyboard
60GB G.Skill Sniper + 2x1TB Spinpoint F3 Raid0 Win 7 Pro x64 ASUS VW266H Razer Blackwidow 
PowerCaseMouse
Seasonic X750 Gold Corsair Carbide 500R White G9 
  hide details  
Reply
ShadowForge
(12 items)
 
Defiant
(14 items)
 
CarbonCat
(13 items)
 
CPUMotherboardGraphicsRAM
Phenom II x6 1405T (unlocked Athlon II X4 640T) ASUS M5A99X EVO AM3+ Asus ENGTX470/2DI/1280MD5/V2 16 GB (4x4GB) G.Skill DDR3 1600 CAS9 1.35v 
Hard DriveHard DriveOptical DriveCooling
WD Caviar Blue 250 2.5" Laptop Drive WD Caviar Black 1TB 3.5" Pioneer BDR-203 BluRay Burner Corsair H60 push 
OSOSMonitorPower
Windows 7 Pro x64 Ubuntu 11.10 Samsung 40" 60hz  ANTEC NEO ECO 520W 
CaseMouse
NZXT Gamma Microsoft Bluetooth Notebook Mouse 5000 
CPUMotherboardGraphicsRAM
i7 - 2600k [5.0 1.42v] ASUS P8Z68 Deluxe Sapphire HD6950 2gb Dirt 3 Edition 8GB G.Skill DDR3 2133 CAS11 
Hard DriveOSMonitorKeyboard
60GB G.Skill Sniper + 2x1TB Spinpoint F3 Raid0 Win 7 Pro x64 ASUS VW266H Razer Blackwidow 
PowerCaseMouse
Seasonic X750 Gold Corsair Carbide 500R White G9 
  hide details  
Reply
post #14 of 18
Quote:
Originally Posted by selectstriker2 View Post
IIRC it wasn't a CPU but 1 or a couple transistors
True.
    
CPUMotherboardGraphicsRAM
Xeon w3680 Gigabyte X58A-UD3R GTX 295 Engineering Samples 2X Corsair XMS3 
Hard DriveHard DriveCoolingOS
Sandisk SSD Seagate Corsair H50 W7 Ult 
MonitorMonitorKeyboardCase
Viewsonic VA1912wb Viewsonic VA912b K350 CM HAF 
Mouse
Logitech M570 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Xeon w3680 Gigabyte X58A-UD3R GTX 295 Engineering Samples 2X Corsair XMS3 
Hard DriveHard DriveCoolingOS
Sandisk SSD Seagate Corsair H50 W7 Ult 
MonitorMonitorKeyboardCase
Viewsonic VA1912wb Viewsonic VA912b K350 CM HAF 
Mouse
Logitech M570 
  hide details  
Reply
post #15 of 18
While it might look like we're "stuck" at 3-4GHz, clock speed does not equate to performance.

Regardless, you can only open and close gates made out of a material so fast before you end up arcing across them instead of activating/deactivating them--this is due to the material being used.

It just so happens that silicon had/has unique properties which suited it well for electronic circuits and microchips, but you can't get around the physics that limit its properties.

It's kinda like engines--you can only get so much horsepower gains from a 1.8L engine (turbo/supercharge, compression ratio, tweaks, etc.) before you hit the realistic limit of that architecture. It's just that we can tweak the current architecture to get more performance out of the same "thing" (maybe think of it like different gearboxes, etc. for the engine analogy), which is what we've been doing with silicon and various architectures for decades.

Related, a 4GHz Pentium-4 does not have nearly the same performance as a 4GHz Sandy Bridge processor.
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
post #16 of 18
Quote:
Originally Posted by Ubernoobie View Post
clock speed=/= performance. The IPC counts too. It takes a amd quad core at 4 to match a 3.2 sb
I like the way you high light the Intel side of it ?

Can you also remember back to the Amd Athlon's 3500+ single cores when an Amd Cpu running at 2200Ghz would beat an Intel running above 3500Ghz and in some instances a dual cored Intel,

Why they do not go above a certain speed all remains in the manufacturing process but then its also down to the companies,

I can remember when they where talking about building a single core CPU with a base clock of 4500Ghz but that was in the days before 92nm,

I am quite happy with them residing in the 3 > 4 GHz margin because for sure the higher the clocks become the higher the price will become to,
RIG 1
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX 8350 ASUS CH-V 990FX EVGA GTX 980 G-SKILL F3-14900 SNIPER 
Hard DriveHard DriveOptical DriveCooling
Samsung 840 Pro WD Black² PIONEER BDR-205 XSPC Raystorm 
OSMonitorKeyboardPower
WIN 8.1 MCE Nec 463 MX5500 DARK POWER PRO 1200 - P10 
CaseMouseMouse PadAudio
CoolerMaster Cosmos II REVO DIRTY ASUS HDAV 1.3 Deluxe 
  hide details  
Reply
RIG 1
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX 8350 ASUS CH-V 990FX EVGA GTX 980 G-SKILL F3-14900 SNIPER 
Hard DriveHard DriveOptical DriveCooling
Samsung 840 Pro WD Black² PIONEER BDR-205 XSPC Raystorm 
OSMonitorKeyboardPower
WIN 8.1 MCE Nec 463 MX5500 DARK POWER PRO 1200 - P10 
CaseMouseMouse PadAudio
CoolerMaster Cosmos II REVO DIRTY ASUS HDAV 1.3 Deluxe 
  hide details  
Reply
post #17 of 18
The reason why we have decided to stay at the 3-4ghz range isn't so much of the fact that we are stuck, but more of the more modern theories that we believe in pushing our efforts to improve processor performance. Back in the Pentium 4 days the gigahertz myth term was used, towards as performance. Once we made a 3.8 ghz, which I think was our limit for the fastest stock speed processor to be released in that generation, we hit a limit. The limit was due to how many transistors we could fit in our cpu and allocate it to the cpu while having enough cache and memory to support the cpu without bottleneck.

We have now gone through many other generations with smaller nanometers and if we wanted to we very well could push through that limit by a decent amount its just the fact that pushing more transistors to a single core would have lower successful yields(successful % of manufactured cpu's) and cost exponentially more to get the same performance that we get with our multi-core direction today.

The benefit is when we shrunk our nm size it was much easier to gain more performance by putting two processors in at the same time and allocating the rest to memory communication between the two, over having one core being pushed to its max and having lower successful manufacturing yeilds in the manufacturing process, resulting in higher costs for making a successful processor, as well as having less multi-tasking capabilities.

Also random note as well we have rapidly improved our memory architecture, and other arithmetical operations with more modern architectures for our gates that has improved our performance while still holding similar gigahertz cpu specs.

The multi-core route is always going to be better then the single core route as each time we lower the nm we are more capable of adding on more cores while keeping the same 3-4ghz range and successful yield rate. The reason we keep the 3-4ghz and not improve it is just simply because its the sweet spot for manufacturing to be able to produce each core with the highest performance, highest yields, while being at the most affordable price.

The problem today is as we add more and more processors we need more and more levels of memory to help the communication between the processors, to prevent the cores from being bottlenecked. This isn't an architecture issue, but an issue just like the gigahertz myth that their is a limit to silicon that if it gets too small instability occurs and as we push ourselves farther to that limit it becomes tougher to manufacture at a reasonable price and eventually might hit silicons physic's limits. So the solution is we switch to other material. The problem though is the solution, cpu manufacturers have not invested nor found a really good alternative yet.

Overall we can push past 3-4ghz but intel limits it because we would get less performance per buck, it would cost more to successfully make a single core as fast as our multi-cores as the yeilds to do so will exponentially drop. The only other alternative would be to expand the size of the cpu's like how gpu's are doing today, the problem is a long time ago the cpu size was set as the larger a cpu gets the more expensive the cpu is just because your buying more material but also because once again yields drop exponentially as you push for larger cpus. So the larger the cpu, the price rises exponentially again. Intel has selected this size once again as its the best way to manufacture the cpu's for the best price points. If you think intel is incapable of blowing through these speeds your wrong its just the fact that the cpu price would rise exponentially and this so far has been the best balance of all the limitations.
Edited by julabask - 10/16/11 at 1:01pm
my system
(13 items)
 
  
CPUMotherboardGraphicsRAM
intel q9450 @ 3.72 ghz evga 750i a1 ftw two evga superclocked gtx 470's in sli gskill 4gb ddr2 1200 5-5-5-15-2T 
Hard DriveOptical DriveOSMonitor
2x western digital black 1tb raid 0 random dvd-rw/cd-rw 22x which somehow uses ata windows 7 64 + linux ubuntu spectre 24inch 
PowerCase
silverstone op1000-e (1000 watts) Silverstone Raven 2 
  hide details  
Reply
my system
(13 items)
 
  
CPUMotherboardGraphicsRAM
intel q9450 @ 3.72 ghz evga 750i a1 ftw two evga superclocked gtx 470's in sli gskill 4gb ddr2 1200 5-5-5-15-2T 
Hard DriveOptical DriveOSMonitor
2x western digital black 1tb raid 0 random dvd-rw/cd-rw 22x which somehow uses ata windows 7 64 + linux ubuntu spectre 24inch 
PowerCase
silverstone op1000-e (1000 watts) Silverstone Raven 2 
  hide details  
Reply
post #18 of 18
Quote:
Originally Posted by julabask View Post
...snip
Carriage return...use it...
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i5-2500K Biostar TP67B+ XFX HD5750 1GB 2x4GB DDR3 Corsair 1600 
Hard DriveOSMonitorPower
60GB OCZ SSD, 2x160GB HDD RAID0, 500GB+500GB+1.5TB Windows 7 Ultimate 64-bit Samsung SyncMaster 930B Antec SmartPower 450w 
Case
Antec 900 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: General Processor Discussions
Overclock.net › Forums › General Hardware › General Processor Discussions › Why are we stuck at 3-4 ghz ???