Overclock.net › Forums › General Hardware › General Processor Discussions › amd vs intel confused still
New Posts  All Forums:Forum Nav:

amd vs intel confused still - Page 9

post #81 of 99
Quote:
Originally Posted by PsyM4n View Post

Quote:
Originally Posted by CynicalUnicorn View Post

RAM is RAM for all intents and purposes. Eight core Jaguar CPU and Pitcairn GPU? That's a PC.

I never said otherwise.

But here we have something with unified memory access shared between the cpu and gpu and short buses (ps4), and something that has non-uniform memory access and long buses, with in-between stop-gaps (amd fx based system).

No matter how you look at it, this is quite the difference.

PC has something like that in the form of AMD APUs.


Now, while it's true that I can do errors/mistakes more often than not, I made sure that I phrased what I had in mind quite properly in this thread so far.

But from the way some things were phrased so far it's quite evident that some of the other ones participating lack proper understanding of how some of the mentioned technologies work, and if/how they affect each other.



Of course bumping into an AMD VS Intel thread is like asking for trouble. But this doesn't give people the right to send insults right in my face. Out of the few threads in which I participated so far, I had the same treatment in like half of them.

Well thing is go into a thread open minded and willing to learn Mantle / DX12 remove alot of those gaps and stops hince term breaks rules and previous barriers look at what Mantle does vs DX in general and how it changes things then come back to this thread with that knowledge and maybe then you could understand why things are so different from previous knowledge / understanding.

Mantle and DX12 change the rules of how things function and take alot of the delays and stops out of the route to do a task.

oh and unified memory yes that plays a role ( Xbox1 doesn't have Unified memory technically has ESRAM and DDR3 ram yet still will be able to use the GPU/ APU's compute power to offset the CPU ) but with the amount of bandwidth possible by the PCIe lanes with PCIe 3.0 x16 and R9 290X 512bit interface and 4gb of high speed DDR5 or even R9 295 x2 which has 2x 4gb of high speed DDR5 do the math that counts as alot of speed for partial usage as computational bandwidth usage of ram on R9 290X and so on. PLZ remember previous consoles had like 256-512mb of ram so section off 1gb or even less for added computational power you do the math what the potential is. A GPU can do alot of the tasks the CPU currently does 100's of times faster with far less performance impact in otherwards the GPU will shrug it off like it is nothing ( 2%-4% GPU usage to do computational work ) while still doing its normal tasks.

you can compare the above stuff to Nvidia GTX Titan and 780TI's and Titan black or Titan Z DX12 will bring this to all of them whether it is Intel / Nvidia or AMD Radeon that is the point of Microsoft taking up Mantle style of API with low level access and a form of Direct Compute using the GPU / APU / IGPU.

here is a link about Xbox one and it not having Unified memory http://www.eurogamer.net/articles/digitalfoundry-xbox-one-memory-better-in-production-hardware so your chat about Unified Memory comparison is lost add to it DX12 on Xbox one bring the GPU doing Direct Compute style computations that CPU currently does in general so do your research before you blurt stuff once again. Xbox one is ESRAM and DDR3 ram none UNIFIED yet still will have the GPU direct compute ability to offload CPU work I linked it directly from Microsoft them saying so.
Edited by rickcooperjr - 4/13/14 at 10:21am
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
post #82 of 99
Quote:
Originally Posted by rickcooperjr View Post

Well thing is go into a thread open minded and willing to learn Mantle / DX12 remove alot of those gaps and stops hince term breaks rules and previous barriers look at what Mantle does vs DX in general and how it changes things then come back to this thread with that knowledge and maybe then you could understand why things are so different from previous knowledge / understanding.

Mantle and DX12 change the rules of how things function and take alot of the delays and stops out of the route to do a task.

They can't change physical rules. You can't make a printed circuit shorter and wider using mantle/dx12. You can't skip fundamental parts of a design just by software. You can't make your PCIe GPU talk directly to the system RAM if it isn't physically possible.

Now, it's quite evident that mantle gives speed boosts, especially on highly specialized processors like APUs. But you cannot yet know how it does this on a low level since the specifications are not available to the public yet. The same limited knowledge applies to dx12 for it's not released yet.

You can write bunches and bunches of paragraphs, but it won't change the lack of low level information on the subject.
post #83 of 99
Quote:
Originally Posted by PsyM4n View Post

Quote:
Originally Posted by rickcooperjr View Post

Well thing is go into a thread open minded and willing to learn Mantle / DX12 remove alot of those gaps and stops hince term breaks rules and previous barriers look at what Mantle does vs DX in general and how it changes things then come back to this thread with that knowledge and maybe then you could understand why things are so different from previous knowledge / understanding.

Mantle and DX12 change the rules of how things function and take alot of the delays and stops out of the route to do a task.

They can't change physical rules. You can't make a printed circuit shorter and wider using mantle/dx12. You can't skip fundamental parts of a design just by software. You can't make your PCIe GPU talk directly to the system RAM if it isn't physically possible.

Now, it's quite evident that mantle gives speed boosts, especially on highly specialized processors like APUs. But you cannot yet know how it does this on a low level since the specifications are not available to the public yet. The same limited knowledge applies to dx12 for it's not released yet.

You can write bunches and bunches of paragraphs, but it won't change the lack of low level information on the subject.

Did you not read the link about DDR4 http://www.micron.com/products/dram/ddr3-to-ddr4 specifically Faster burst accesses / Migration to higher‐speed I/O / More banks these eliminate alot of the added length of other paths needed for the ram on the PCB in otherwards cuts the length of wire / time needed to move signal across board add to it system ram doesn't matter they said will use the High speed DDR5 ram on the GPU to do these tasks rather than system ram because of the much faster speeds and also bypass the CPU memory controller issues you just mentioned.

PLZ read my posts before you blurt out exact opposite of what I typed.
Edited by rickcooperjr - 4/13/14 at 10:04am
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
post #84 of 99
Quote:
Originally Posted by rickcooperjr View Post

Did you not read the link about DDR4 http://www.micron.com/products/dram/ddr3-to-ddr4 specifically Faster burst accesses / Migration to higher‐speed I/O / More banks these eliminate alot of the added length of other paths needed for the ram on the PCB in otherwards cuts the length of wire / time needed to move signal across board add to it system ram doesn't matter they said will use the High speed DDR5 ram on the GPU to do these tasks rather than system ram because of the much faster speeds and also bypass the CPU memory controller issues you just mentioned.

PLZ read my posts before you blurt out exact opposite of what I typed.

How exactly do those take out the limits on systems using the FX processors you mentioned earlier?

Even if we do take that out and use a desktop that uses ddr4, a PCIe GPU will still need to send data to the system RAM through the PCIe bus and the CPU.
post #85 of 99
Quote:
Originally Posted by PsyM4n View Post

Quote:
Originally Posted by rickcooperjr View Post

Did you not read the link about DDR4 http://www.micron.com/products/dram/ddr3-to-ddr4 specifically Faster burst accesses / Migration to higher‐speed I/O / More banks these eliminate alot of the added length of other paths needed for the ram on the PCB in otherwards cuts the length of wire / time needed to move signal across board add to it system ram doesn't matter they said will use the High speed DDR5 ram on the GPU to do these tasks rather than system ram because of the much faster speeds and also bypass the CPU memory controller issues you just mentioned.

PLZ read my posts before you blurt out exact opposite of what I typed.

How exactly do those take out the limits on systems using the FX processors you mentioned earlier?

Even if we do take that out and use a desktop that uses ddr4, a PCIe GPU will still need to send data to the system RAM through the PCIe bus and the CPU.

I added this to a previous post to answer these questions do your research and home work

here is a link about Xbox one and it not having Unified memory http://www.eurogamer.net/articles/digitalfoundry-xbox-one-memory-better-in-production-hardware so your chat about Unified Memory comparison is lost add to it DX12 on Xbox one bring the GPU doing Direct Compute style computations that CPU currently does in general so do your research before you blurt stuff once again. Xbox one is ESRAM and DDR3 ram none UNIFIED yet still will have the GPU direct compute ability to offload CPU work I linked it directly from Microsoft them saying so.

all of these factors take your before mentioned rules and limitations out the windows otherwise microsoft would never have said this was coming with Xbox one and the DX12 for it later on. So if you want to argue with Microsoft I believe they are way more educated and knowledgable about these things especially since its theyre baby / hardware and tech and even programming ( hince DX12 ) they will be using they already answered all your questions and disproved all your facts you threw out do some web searching and reading and maybe emailing Microsoft asking questions maybe you will get a bit more knowledge this way.

this is specifically the one you need to read short and to the point http://www.theverge.com/gaming/2014/3/20/5530420/microsoft-claims-directx-12-will-improve-xbox-one-games-wont-require
Edited by rickcooperjr - 4/13/14 at 10:34am
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
post #86 of 99
Quote:
Originally Posted by rickcooperjr View Post

I added this to a previous post to answer these questions do your research and home work

here is a link about Xbox one and it not having Unified memory http://www.eurogamer.net/articles/digitalfoundry-xbox-one-memory-better-in-production-hardware so your chat about Unified Memory comparison is lost add to it DX12 on Xbox one bring the GPU doing Direct Compute style computations that CPU currently does in general so do your research before you blurt stuff once again. Xbox one is ESRAM and DDR3 ram none UNIFIED yet still will have the GPU direct compute ability to offload CPU work I linked it directly from Microsoft them saying so.

Right back at you. You're essentially talking about a bunch of different things that you don't seem to know of, nor the relation they have.

Both the XB360 and the XB1 use unified RAM. It's part of their design. Even something as old as the Nintendo 64 did use unified RAM.
post #87 of 99
Quote:
Originally Posted by PsyM4n View Post

Quote:
Originally Posted by rickcooperjr View Post

I added this to a previous post to answer these questions do your research and home work

here is a link about Xbox one and it not having Unified memory http://www.eurogamer.net/articles/digitalfoundry-xbox-one-memory-better-in-production-hardware so your chat about Unified Memory comparison is lost add to it DX12 on Xbox one bring the GPU doing Direct Compute style computations that CPU currently does in general so do your research before you blurt stuff once again. Xbox one is ESRAM and DDR3 ram none UNIFIED yet still will have the GPU direct compute ability to offload CPU work I linked it directly from Microsoft them saying so.

Right back at you. You're essentially talking about a bunch of different things that you don't seem to know of, nor the relation they have.

Both the XB360 and the XB1 use unified RAM. It's part of their design. Even something as old as the Nintendo 64 did use unified RAM.

wrong they dont use unified memory on XB1 I already linked proof of this they use ESRAM and DDR3 ram not fully unified and the ram is partitioned off by the OS meening seperated again not unified. XB1 uses a workaround similar but not truly unified and is partitioned off by the OS itself. Like a 2 partioned harddrive functiones in general like 2 hardrives but truly is a single drive but they both share read / write bandwidths yet have to wait on each other to do one or the others task.
Edited by rickcooperjr - 4/13/14 at 10:42am
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
post #88 of 99
Quote:
Originally Posted by rickcooperjr View Post

wrong they dont use unified memory on XB1 I already linked proof of this they use ESRAM and DDR3 ram not fully unified and the ram is partitioned off by the OS meening seperated again not unified. XB1 uses a workaround similar but not truly unified and is partitioned off by the OS itself. Like a 2 partioned harddrive functiones in general like 2 hardrives but truly is a single drive but they both share read / write bandwidths yet have to wait on each other to do one or the others task.

Seriously you need to take a look at the XBOX One datasheet or something. A course of memory designs would be a plus too. The ESRAM is there to speed things up cause unified ddr3 is not exactly fast by today's standards.
post #89 of 99
Quote:
Originally Posted by PsyM4n View Post

Quote:
Originally Posted by rickcooperjr View Post

wrong they dont use unified memory on XB1 I already linked proof of this they use ESRAM and DDR3 ram not fully unified and the ram is partitioned off by the OS meening seperated again not unified. XB1 uses a workaround similar but not truly unified and is partitioned off by the OS itself. Like a 2 partioned harddrive functiones in general like 2 hardrives but truly is a single drive but they both share read / write bandwidths yet have to wait on each other to do one or the others task.

Seriously you need to take a look at the XBOX One datasheet or something. A course of memory designs would be a plus too. The ESRAM is there to speed things up cause unified ddr3 is not exactly fast by today's standards.

Facts are the way they use ESRAM is exactly how they could use some of the GPU high speed DDR5 ram to accelerate things the PCIe bandwidth and 512bit interface also makes a perfect match up for this do you not understand you answered your own contradiction or they could simply add an extra Ram chip on the PCB of the mobo or GPU dedicated to do this exact task simple fix and problem solved just as Microsoft did with the ESRAM.
Edited by rickcooperjr - 4/13/14 at 11:09am
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel I7 5820k 4.6ghz core 4.3ghz CPU cache 1.345v asus sabertooth X99 sold R9 290x's am currently running a R9 280A X... 16gb of ddr4 corsair vengeance LPX 3200mhz 1t 1... 
Hard DriveOptical DriveCoolingOS
M.2 Toshiba OCZ RD400 512gb ASUS DVD rw  liquid cooled cpu and gpu's windows 10 pro 64bit 
MonitorKeyboardPowerCase
3 x 32lcd/led 2560x1600 wolfking timberwolf EVGA supernova 1300w G2  Thermaltake core X9 
MouseMouse PadOther
steel series wow cataclysm 14button programable  wolfking sniper Intel PRO PT dual port PCIe server NIC 
  hide details  
Reply
post #90 of 99
Quote:
Originally Posted by rickcooperjr View Post

Facts are the way they use ESRAM is exactly how they could use some of the GPU high speed DDR5 ram to accelerate things the PCIe bandwidth and 512bit interface also makes a perfect match up for this do you not understand you answered your own contradiction or they could simply add an extra Ram chip on the PCB of the mobo or GPU dedicated to do this exact task simple fix and problem solved just as Microsoft did with the ESRAM.

Seriously, write less, read more. We can talk again after that. smile.gif
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: General Processor Discussions
Overclock.net › Forums › General Hardware › General Processor Discussions › amd vs intel confused still