Overclock.net › Forums › Industry News › Hardware News › [Softpedia] AMD Trinity Architectural Preview
New Posts  All Forums:Forum Nav:

[Softpedia] AMD Trinity Architectural Preview - Page 5

post #41 of 46
Quote:
Originally Posted by sergionography View Post

how in the world did you come up with this conclusion? how is 3 seconds come out to be 3 %?
3 seconds short from 20 is 3/20 which is 15/100 = 15%

Quote:
Originally Posted by sumitlian View Post

its 15% bro, if the above bench you have showed is true then Llano is 15% faster than trinity.

These are not theoretical results, just like AMD slides. If you want you can break them down to actual percentage increases.
Main Rig
(14 items)
 
Linux Rig
(10 items)
 
 
CPUMotherboardGraphicsRAM
Athlon II x3 450 Biostar A880GZ PowerColor HD 4650 DDR3 4GB Samsung MV-3V2G3/US 
Hard DriveOptical DriveCoolingOS
Samsung 830 128GB (Pending) ASUS DRW-24B1ST Stock Windows 8 Enterprise x64 
KeyboardPowerCaseMouse
Logitech Classic 200 300w Hipro a1410n HP 1000 DPI 
Mouse PadAudio
Standard Realtek ALC662 
CPUMotherboardGraphicsRAM
AMD Sempron 3000+ ASUS A7V8X-LA VIA KM400A 2GB DDR 333 
Hard DriveOptical DriveCoolingOS
40GB Seagate Barracuda 7200.7 Samsung Writemaster Stock ASUS Ubuntu 12.04 
PowerCase
Hipro 250W SR1500NX 
  hide details  
Reply
Main Rig
(14 items)
 
Linux Rig
(10 items)
 
 
CPUMotherboardGraphicsRAM
Athlon II x3 450 Biostar A880GZ PowerColor HD 4650 DDR3 4GB Samsung MV-3V2G3/US 
Hard DriveOptical DriveCoolingOS
Samsung 830 128GB (Pending) ASUS DRW-24B1ST Stock Windows 8 Enterprise x64 
KeyboardPowerCaseMouse
Logitech Classic 200 300w Hipro a1410n HP 1000 DPI 
Mouse PadAudio
Standard Realtek ALC662 
CPUMotherboardGraphicsRAM
AMD Sempron 3000+ ASUS A7V8X-LA VIA KM400A 2GB DDR 333 
Hard DriveOptical DriveCoolingOS
40GB Seagate Barracuda 7200.7 Samsung Writemaster Stock ASUS Ubuntu 12.04 
PowerCase
Hipro 250W SR1500NX 
  hide details  
Reply
post #42 of 46
Quote:
Originally Posted by fantasyalive View Post

Llano uses the stars core which is considered to have a 6% higher IPC then Phenom II, while bulldozer cores are considered to have 11% decrease in IPC.

Please note that "Stars" is a family of cores. It's the naming used for desktop products based on the same architecture (K10.5) such as Deneb, Propus, Thuban, etc. These are the cores that make up the various Phenom II flavors. The implication is that "stars" cores would be equal to Phenom II performance because they are the very cores that make them up.
post #43 of 46
Quote:
Originally Posted by Particle View Post

Please note that "Stars" is a family of cores. It's the naming used for desktop products based on the same architecture (K10.5) such as Deneb, Propus, Thuban, etc. These are the cores that make up the various Phenom II flavors. The implication is that "stars" cores would be equal to Phenom II performance because they are the very cores that make them up.

i think he meant the HUSKY cores in llano are 6% faster
Quote:
Originally Posted by Warmonger View Post

These are not theoretical results, just like AMD slides. If you want you can break them down to actual percentage increases.

well you did state that ttrinity is 3% slower than llanno clock -clock and then posted the superpi m results thats why i got confused
because if llano took 20 seconds and trinity took it 3 extra seconds at 23 to finish the calculation then it finished 15% behind
but it was already established on this thread that superpi is an outdated benchmark using old x87 instruction which isnt supported anymore
post #44 of 46
Quote:
Originally Posted by sergionography View Post

i think he meant the HUSKY cores in llano are 6% faster
well you did state that ttrinity is 3% slower than llanno clock -clock and then posted the superpi m results thats why i got confused
because if llano took 20 seconds and trinity took it 3 extra seconds at 23 to finish the calculation then it finished 15% behind
but it was already established on this thread that superpi is an outdated benchmark using old x87 instruction which isnt supported anymore

Its still a suitable benchmark, and always will be. Both processors dont support the instruction so its impossible for one to have an advantage over the other. Instead it relies heavily on the processors floating point unit to make these calculations. Which results in how fast the processor can do math, tho I still don't get how math calculations are a direct result to overall processor performance. headscratch.gif
Main Rig
(14 items)
 
Linux Rig
(10 items)
 
 
CPUMotherboardGraphicsRAM
Athlon II x3 450 Biostar A880GZ PowerColor HD 4650 DDR3 4GB Samsung MV-3V2G3/US 
Hard DriveOptical DriveCoolingOS
Samsung 830 128GB (Pending) ASUS DRW-24B1ST Stock Windows 8 Enterprise x64 
KeyboardPowerCaseMouse
Logitech Classic 200 300w Hipro a1410n HP 1000 DPI 
Mouse PadAudio
Standard Realtek ALC662 
CPUMotherboardGraphicsRAM
AMD Sempron 3000+ ASUS A7V8X-LA VIA KM400A 2GB DDR 333 
Hard DriveOptical DriveCoolingOS
40GB Seagate Barracuda 7200.7 Samsung Writemaster Stock ASUS Ubuntu 12.04 
PowerCase
Hipro 250W SR1500NX 
  hide details  
Reply
Main Rig
(14 items)
 
Linux Rig
(10 items)
 
 
CPUMotherboardGraphicsRAM
Athlon II x3 450 Biostar A880GZ PowerColor HD 4650 DDR3 4GB Samsung MV-3V2G3/US 
Hard DriveOptical DriveCoolingOS
Samsung 830 128GB (Pending) ASUS DRW-24B1ST Stock Windows 8 Enterprise x64 
KeyboardPowerCaseMouse
Logitech Classic 200 300w Hipro a1410n HP 1000 DPI 
Mouse PadAudio
Standard Realtek ALC662 
CPUMotherboardGraphicsRAM
AMD Sempron 3000+ ASUS A7V8X-LA VIA KM400A 2GB DDR 333 
Hard DriveOptical DriveCoolingOS
40GB Seagate Barracuda 7200.7 Samsung Writemaster Stock ASUS Ubuntu 12.04 
PowerCase
Hipro 250W SR1500NX 
  hide details  
Reply
post #45 of 46
Quote:
Originally Posted by Warmonger View Post

Its still a suitable benchmark, and always will be. Both processors dont support the instruction so its impossible for one to have an advantage over the other. Instead it relies heavily on the processors floating point unit to make these calculations. Which results in how fast the processor can do math, tho I still don't get how math calculations are a direct result to overall processor performance. headscratch.gif

Not really, cause then you don't know how much they worked on supported instructions. Maybe they increased those quite a bit, making the new cores much better. This is the reason why benchmark programs should be designed for the Arch/CPU/GPU.
Current Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 4.6GHz@1.44v GA-990FXA-UD3 R4.0 HD 7950 (1100/1450) 8G Muskin DDR3 1866@8CLS 
Hard DriveOptical DriveOSMonitor
1TB WD LiteOn DVD-RW DL Linux/Windows 19" Phillips TV 1080p 
PowerCaseMouseMouse Pad
OCZ 600W Generic Junk Logitech MX400 Generic Junk 
Audio
SBL 5.1 
  hide details  
Reply
Current Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 4.6GHz@1.44v GA-990FXA-UD3 R4.0 HD 7950 (1100/1450) 8G Muskin DDR3 1866@8CLS 
Hard DriveOptical DriveOSMonitor
1TB WD LiteOn DVD-RW DL Linux/Windows 19" Phillips TV 1080p 
PowerCaseMouseMouse Pad
OCZ 600W Generic Junk Logitech MX400 Generic Junk 
Audio
SBL 5.1 
  hide details  
Reply
post #46 of 46
Quote:
Originally Posted by mushroomboy View Post

Not really, cause then you don't know how much they worked on supported instructions. Maybe they increased those quite a bit, making the new cores much better. This is the reason why benchmark programs should be designed for the Arch/CPU/GPU.

exactly, one has to consider that at the end of the day companies like intel, amd, arm, nvidia and the rest of the hardware manufactures all make hardware to create new possibilities for software and what not, it is up to developers whether they want to take full advantage or not, for example a software optimized with opencl might run way better on an amd graphic card than on nvidia, but with cuda its the other way around, and software done according to intel compilers will run typically better on intel than on amd,
the problem with these benchmarks and software as a whole today is that their all pretty much done according to intels standards/compilers/approved instruction sets, so nomatter what aslong as this is the case amd will keep playing catchup, unless they introduce something better as a platform, and even if they do it will be hard to push it and make it a standard since amd doesnt have the money and resources to do so
and in my opinion i think it is intel pretty much holding the industry back in many cases, recently i heard news about intel demanding higher resolution standards from OEM's around 2013-2014 when haswell is around, has it been up 2 amd such standards wouldve been pushed a long time ago since amd has much better graphics capability but offcourse only when intel has the hardware to support will the industry ever change -__-
as much as ipc and fat cores are important they really dont change anything to the game, at the end of the day performance gains as a whole are gained through newer instruction sets, more efficient ways of execution, accelerated software, and so on so forth, and as far as i know intel did very little to improve things and as instruction sets they were used only for business purposes to keep amd out.
i just hope amd can pull something great with their HSA and upcomming compilers that they were speaking about and turn the tables around a bit. that is exactly why they bought ati at the first place, to have an advantage and capitalize on intels weakness, which gained them 46% of the desktop market so far with their apu's, lets just hope trinity keeps that tradition and helps them gain more ground
Edited by sergionography - 5/10/12 at 12:28am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [Softpedia] AMD Trinity Architectural Preview