Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [WCCF] AMD’s New High Performance Processor Cores Coming Sometime in 2015 – Giving Up on Modular Architecture
New Posts  All Forums:Forum Nav:

[WCCF] AMD’s New High Performance Processor Cores Coming Sometime in 2015 – Giving Up on Modular Architecture - Page 33

post #321 of 485
Quote:
Originally Posted by sdlvx View Post

To me, SMT is one of those things that sounds good on paper and then in real world it's not so good.

It shows you haven't used an i7. The difference between 4 cores and 8 virtual cores on heavy and unleashed multithreading is immense. It's not an illusion either since the same chip without HT will be the same speed when HT is disabled on the HT chip (factoring out variations in total cache size).

The main confusion comes from people thinking those are virtual cores. No better than software cores. That's the illusion of the masses.

They are not. They're partly SMP hardware. In general though I bet both AMD and Intel are good designers and it goes down to foundry.
Edited by fateswarm - 5/10/14 at 8:54am
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
post #322 of 485
Quote:
Originally Posted by fateswarm View Post

Quote:
Originally Posted by sdlvx View Post

To me, SMT is one of those things that sounds good on paper and then in real world it's not so good.

It shows you haven't used an i7. The difference between 4 cores and 8 virtual cores on heavy and unleashed multithreading is immense. It's not an illusion either since the same chip without HT will be the same speed when HT is disabled on the HT chip (factoring out variations in total cache size).

The main confusion comes from people thinking those are virtual cores. No better than software cores. That's the illusion of the masses.

They are not. They partially SMP hardware. In general though I bet both AMD and Intel are good designers and it goes down to foundry.

I have an i7 920 sitting right next to me and it's installing Gentoo (still, had some blocking packages so it's taking a while). I hand optimized the kernel and have the best GCC settings I can get for the chip. In fact, I had that i7 920 for several years, I got it at launch and had it until Piledriver came out. So i actually used i7 920 for nearly 4 years. And before that I had Pentium 4s with HT for another 4 years or so.

So I realize you are thinking that I don't know what I'm talking about but I do. HT on has always given me a system that is somewhat faster and not usable when all the cores are under 100% load. I have around 8 years of experience working with Intel SMT/Hyperthreading.

I don't like it and it's always been a sort of "well I have to use it for these tasks because it makes certain things I'm doing faster but my overall experience is going to be degraded because the system won't be as responsive."

And as I said before, When i7 920 came out, the Windows scheduler in XP was not up to the task and loved to schedule threads completely wrong. It is the same issue that AMD deals with with modules, except Intel more than likely just threw a lot of money at Windows and said "make it better!"
Tyrant Kuma
(13 items)
 
Starscythe
(13 items)
 
Mobility
(6 items)
 
CPUCPUCPUCPU
Opteron 8431 Opteron 8431 Opteron 8431 Opteron 8431 
MotherboardGraphicsRAMHard Drive
Super Micro h8qme-2+ Sapphire 4870 Samsung DDR2 ECC 667mhz western digital caviar blue 
CoolingCoolingOSPower
FX 8350 stock cooler FX 6300 stock cooler Gentoo Linux 2x Silverstone PSU 
Case
Custom Fabbed Steel Case 
CPUMotherboardGraphicsRAM
AMD FX 8350 gigabyte 990FXA-UD5 7970 Mushkin Enhanced Blackline 16GB 
OSOSMonitorKeyboard
Gentoo Linux Windows 7 x64 Yamakasi Catleap Q270 Corsair K90 
PowerCaseMouseAudio
Silverstone Strider Gold Evolution 850W 80 PLUS... Antec 1200 Logitech G9x Asus Xonar D2X 
Other
XSPC RS360 Raystorm with custom radiator mounts 
CPUGraphicsRAMOS
a4 5000 Radeon HD 8330 8GB G. Skill DDR3 1600 9-9-9 Gentoo Linux 
OSOther
Windows 7 Lenovo Thinkpad x140e 
  hide details  
Reply
Tyrant Kuma
(13 items)
 
Starscythe
(13 items)
 
Mobility
(6 items)
 
CPUCPUCPUCPU
Opteron 8431 Opteron 8431 Opteron 8431 Opteron 8431 
MotherboardGraphicsRAMHard Drive
Super Micro h8qme-2+ Sapphire 4870 Samsung DDR2 ECC 667mhz western digital caviar blue 
CoolingCoolingOSPower
FX 8350 stock cooler FX 6300 stock cooler Gentoo Linux 2x Silverstone PSU 
Case
Custom Fabbed Steel Case 
CPUMotherboardGraphicsRAM
AMD FX 8350 gigabyte 990FXA-UD5 7970 Mushkin Enhanced Blackline 16GB 
OSOSMonitorKeyboard
Gentoo Linux Windows 7 x64 Yamakasi Catleap Q270 Corsair K90 
PowerCaseMouseAudio
Silverstone Strider Gold Evolution 850W 80 PLUS... Antec 1200 Logitech G9x Asus Xonar D2X 
Other
XSPC RS360 Raystorm with custom radiator mounts 
CPUGraphicsRAMOS
a4 5000 Radeon HD 8330 8GB G. Skill DDR3 1600 9-9-9 Gentoo Linux 
OSOther
Windows 7 Lenovo Thinkpad x140e 
  hide details  
Reply
post #323 of 485
Quote:
Originally Posted by Seronx View Post

This is a false claim in fact hyperthreading on provides no substantial boosts in today's workloads.

No, what you have posted here is a false claim, and one easily debunked at that.

You offer a lot of irrelevant architectural details, as if they do something to prove that HT cannot work...yet it does. Clearly, you are missing something. You sound like someone explaining the physics of how water cannot be wet to a crowd of people soaked to the skin in a downpour. Most of them don't know exactly what you are saying, but nearly every single one of them knows it's bull regardless.

HT is not always an improvement, and is very rarely a detriment, but most of the time it helps, and an in a significant fraction of workloads, it helps a lot. Yes, even in modern workloads.

Why don't you make a list of common, modern, real-world workloads I can simulate, and I can provide some HT on vs. HT off results?
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #324 of 485
I keep hearing badmouthing for the i7 mainly pre-Sandy Bridge. Did they improve HT considerably after Sandy Bridge? I've only used i7s after its release.
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
post #325 of 485
Quote:
Originally Posted by fateswarm View Post

I keep hearing badmouthing for the i7 mainly pre-Sandy Bridge. Did they improve HT considerably after Sandy Bridge? I've only used i7s after its release.

There have been minor tweaks over time, but even the original Pentium IV Northwood implementation of HT was an asset far more often than not.

Nehalem and Westmere HT work very well, and the bulk of my HT enabled processors pre-date Sandy Bridge.

Can I find something that will produce worse results with HT on? Absolutely. However, if I remove those applications that are not synthetic benchmarks or stress tests, I'm mostly just left with database servers that allocate resources based on logical processor count without taking into account that not all logical cores are physical cores. I don't run many database servers, and I doubt most people on OCN do either.

Most of the problems people mention with HT are either complete farces, or come from those who refuse to make any changes to how they do something to take best advantage of it. If you have a (fairly uncommon) scenario where full load HT use is less responsive than full load non-HT use, why are you not lowering the priority of the application, or reducing the thread count it uses by one? Do this and all of your responsiveness will return, and whatever you are doing will almost certainly still finish appreciably faster than if HT was disabled.

Most programs will still work perfectly fine, and still see tangible benefits, even if nothing is done differently, but for the outliers, spending the extra five friggin seconds it takes to remind them that HT is not exactly the same as adding more cores, can be well worth it.
Edited by Blameless - 5/10/14 at 9:27am
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #326 of 485
Quote:
Originally Posted by fateswarm View Post

I keep hearing badmouthing for the i7 mainly pre-Sandy Bridge. Did they improve HT considerably after Sandy Bridge? I've only used i7s after its release.

If you look at a benchmarketing site you'll see things like "wow amazing 20% faster with HT on with SB!" and if you look at something where performance actually matters, like HPC, they tell you to turn it off (http://www.microway.com/hpc-tech-tips/achieve-the-best-performance-intel-xeon-e5-2600-sandy-bridge/)

I poked around some other forums (I think it's against the TOS to link to them at OCN) but it seems like HT in SB still suffers somewhat from the same issues. HT used to actually hurt performance significantly in gaming and with SB it seems like it's either a small hit or a small gain. But again, I'm seeing people say a lot different things in forums and professional environments as opposed to benchmarking and tech sites. Which is raising some red flags to me.

But it still raises the point that you can definitely botch SMT and make it something that hurts performance and AMD going SMT for their next CPU core could be just as bad (or even worse) than CMT. I am going to have to pick up a used SB/IB/Haswell or something so I can test all this stuff but in the Nehalem and older days it was pretty much accepted on forums that HT was completely useless at best and hurt your performance significantly at worst.

But even if it is fixed 100% in Haswell to the point where it always gave 20% improvement, it took Intel a ton of iterations to make HT not lower your game FPS by 30%, I think it's rather foolish to expect AMD to be able to release a very solid version of SMT on a brand new architecture when it was so difficult for Intel.
Tyrant Kuma
(13 items)
 
Starscythe
(13 items)
 
Mobility
(6 items)
 
CPUCPUCPUCPU
Opteron 8431 Opteron 8431 Opteron 8431 Opteron 8431 
MotherboardGraphicsRAMHard Drive
Super Micro h8qme-2+ Sapphire 4870 Samsung DDR2 ECC 667mhz western digital caviar blue 
CoolingCoolingOSPower
FX 8350 stock cooler FX 6300 stock cooler Gentoo Linux 2x Silverstone PSU 
Case
Custom Fabbed Steel Case 
CPUMotherboardGraphicsRAM
AMD FX 8350 gigabyte 990FXA-UD5 7970 Mushkin Enhanced Blackline 16GB 
OSOSMonitorKeyboard
Gentoo Linux Windows 7 x64 Yamakasi Catleap Q270 Corsair K90 
PowerCaseMouseAudio
Silverstone Strider Gold Evolution 850W 80 PLUS... Antec 1200 Logitech G9x Asus Xonar D2X 
Other
XSPC RS360 Raystorm with custom radiator mounts 
CPUGraphicsRAMOS
a4 5000 Radeon HD 8330 8GB G. Skill DDR3 1600 9-9-9 Gentoo Linux 
OSOther
Windows 7 Lenovo Thinkpad x140e 
  hide details  
Reply
Tyrant Kuma
(13 items)
 
Starscythe
(13 items)
 
Mobility
(6 items)
 
CPUCPUCPUCPU
Opteron 8431 Opteron 8431 Opteron 8431 Opteron 8431 
MotherboardGraphicsRAMHard Drive
Super Micro h8qme-2+ Sapphire 4870 Samsung DDR2 ECC 667mhz western digital caviar blue 
CoolingCoolingOSPower
FX 8350 stock cooler FX 6300 stock cooler Gentoo Linux 2x Silverstone PSU 
Case
Custom Fabbed Steel Case 
CPUMotherboardGraphicsRAM
AMD FX 8350 gigabyte 990FXA-UD5 7970 Mushkin Enhanced Blackline 16GB 
OSOSMonitorKeyboard
Gentoo Linux Windows 7 x64 Yamakasi Catleap Q270 Corsair K90 
PowerCaseMouseAudio
Silverstone Strider Gold Evolution 850W 80 PLUS... Antec 1200 Logitech G9x Asus Xonar D2X 
Other
XSPC RS360 Raystorm with custom radiator mounts 
CPUGraphicsRAMOS
a4 5000 Radeon HD 8330 8GB G. Skill DDR3 1600 9-9-9 Gentoo Linux 
OSOther
Windows 7 Lenovo Thinkpad x140e 
  hide details  
Reply
post #327 of 485
Quote:
Originally Posted by sdlvx View Post

if you look at something where performance actually matters, like HPC, they tell you to turn it off (http://www.microway.com/hpc-tech-tips/achieve-the-best-performance-intel-xeon-e5-2600-sandy-bridge/)

I'd argue that performance actually matters for damn near everyone here, yet almost no one here (with a handful of exceptions) is doing HPC work on their systems.

Even article you posted suggests HT be evaluated on a case by case basis.
Quote:
Originally Posted by sdlvx View Post

HT used to actually hurt performance significantly in gaming

No it didn't.
Quote:
Originally Posted by sdlvx View Post

But again, I'm seeing people say a lot different things in forums and professional environments as opposed to benchmarking and tech sites. Which is raising some red flags to me.

Why would it raise red flags? If you have HT capable parts, you can easily check to see whether it's an advantage or disadvantage for what you are doing. If you are looking at buying a part with HT, you can find tests for what you are or will be doing that compares HT on and off.

How it performs in things you do not do and will not be doing are irrelevant.
Quote:
Originally Posted by sdlvx View Post

in the Nehalem and older days it was pretty much accepted on forums that HT was completely useless at best and hurt your performance significantly at worst.

Utter nonsense, to the point that I am starting to suspect your goal is to deliberately mislead people.
Quote:
Originally Posted by sdlvx View Post

it took Intel a ton of iterations to make HT not lower your game FPS by 30%, I think it's rather foolish to expect AMD to be able to release a very solid version of SMT on a brand new architecture when it was so difficult for Intel.

More nonsense.
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #328 of 485
Either way let's all agree that CMT can be better than HT because that is just the way it is. I hope Jimm and it's team put effort in creating a large SMT core with CMT for things like cache and other things that can be shared without real penalty.

I hope to see Excavator and maybe the unlocked version if it exists and if there is actually anything to be unlocked earlier let's say Q4 2014 would be great. But with how slowly the Kaveri push to mobile and desktop went and still goes I sincerely doubt it will happen. If we see it early 2015 like we did with Kaveri it can of course still be good but if they are completely switching I still think they should just rush it out the door and focus on the new arch.

AMD is in a though spot due to them having to divide their budget for their dev teams over ARM and X86 however it gives them a potential huge advantage especially with their HSA methodology.
post #329 of 485
Quote:
Originally Posted by sdlvx View Post

...AMD going SMT for their next CPU core could be just as bad (or even worse) than CMT. I am going to have to pick up a used SB/IB/Haswell or something so I can test all this stuff but in the Nehalem and older days it was pretty much accepted on forums that HT was completely useless at best and hurt your performance significantly at worst.
It states 2015 which is Excavator.

"AMD’s New High Performance Processor Cores Coming Sometime in 2015"

Also, high performance cores can mean anything like a successor to Warsaw. Warsaw launched in Q1 2014, its successor will launch in Q1 2015. While, on the consumer side there will be an FX version.

To note how SMT is used in SteamrollerB and Excavator, you have to look at the thread mechanisms in Bulldozer and Piledriver.


Bulldozer/Piledriver:
Branch Predict Thread Domain; Vertical Multithreading/Temporal
Instruction Fetch Thread Domain; Vertical Multithreading/Temporal
Decode Thread Domain; Vertical Multithreading/Temporal
Dispatch Thread Domain; Vertical Multithreading/Temporal
Both Cores; Each a single thread
FP Front-end; Vertical Multithreading/Temporal
FP Execution; Simultaneously Multithreading/Thread Agnostic
FP Back-end; is tied to each core's retirement queue/LSU.
L2/CU; Simultaneously Multithreading/Thread Agnostic

Steamroller/Excavator:
Branch Predict Thread Domain; SMT/Thread Agnostic (This is for Single Thread or Dual Thread biasing)
Instruction Fetch Thread Domain; SMT/Thread Agnostic (This is for Single Thread or Dual Thread biasing)
Decode Thread Domain; 2x Single Thread/SMT/Not Thread Agnostic
Dispatch Thread Domain; 2x Single Thread/SMT/Not Thread Agnostic
Both Cores; Each a single thread
FP Front-end; 2x Single Thread/SMT/Not Thread Agnostic
FP Execution; SMT/Thread Agnostic
FP Back-end; is tied to each core's retirement queue/LSU.
L2/CU; SMT/Thread Agnostic

Where Bulldozer/Piledriver for the Front-end and FPU performs like Intel's Hyperthreading, Steamroller/Excavator operates more or less like two CMP cores that have access to each others Front-end and FPU resources.
Edited by Seronx - 5/10/14 at 10:22am
AMD FX ~Seronx
(16 items)
 
  
CPUMotherboardGraphicsRAM
FX-9800P Acer Wasp R7 M440 SK Hynix HMA41GS6AFR8N-TF 
Hard DriveHard DriveOptical DriveCooling
KINGSTON RBU-SNS8152S3128GG2 TOSHIBA MQ01ABD100 HL-DT-ST DVDRAM GUE1N Stock 
OSMonitorKeyboardPower
Microsoft Windows 10 Home Build 14393 Viewsonic XG2401 24 Hz-144 Hz Ducky Channel Shine 3 Stock 65W 
CaseMouseMouse PadAudio
Acer Exoskeleton Steelseries Rival 300 Razer Megasoma AMD-Realtek ALC255 
  hide details  
Reply
AMD FX ~Seronx
(16 items)
 
  
CPUMotherboardGraphicsRAM
FX-9800P Acer Wasp R7 M440 SK Hynix HMA41GS6AFR8N-TF 
Hard DriveHard DriveOptical DriveCooling
KINGSTON RBU-SNS8152S3128GG2 TOSHIBA MQ01ABD100 HL-DT-ST DVDRAM GUE1N Stock 
OSMonitorKeyboardPower
Microsoft Windows 10 Home Build 14393 Viewsonic XG2401 24 Hz-144 Hz Ducky Channel Shine 3 Stock 65W 
CaseMouseMouse PadAudio
Acer Exoskeleton Steelseries Rival 300 Razer Megasoma AMD-Realtek ALC255 
  hide details  
Reply
post #330 of 485
Quote:
Originally Posted by maarten12100 View Post

Either way let's all agree that CMT can be better than HT because that is just the way it is. I hope Jimm and it's team put effort in creating a large SMT core with CMT for things like cache and other things that can be shared without real penalty.

I hope to see Excavator and maybe the unlocked version if it exists and if there is actually anything to be unlocked earlier let's say Q4 2014 would be great. But with how slowly the Kaveri push to mobile and desktop went and still goes I sincerely doubt it will happen. If we see it early 2015 like we did with Kaveri it can of course still be good but if they are completely switching I still think they should just rush it out the door and focus on the new arch.

AMD is in a though spot due to them having to divide their budget for their dev teams over ARM and X86 however it gives them a potential huge advantage especially with their HSA methodology.

 

Irregardless of the entire argument above me, I have to say that I think this is going to be a long term mistaken for AMD. Dividing their development resources is going to put them at a deeper disadvantage than they already are in the market. They don't have unlimited resources to work around with. And they are taking far too large of a gamble. Especially when that gamble is still met with stiff resistance from either the ARM or x86 side. And especially since their new architecture isn't scheduled to come out until 2016, which in business terms is quite a ways away. And in technology terms, is even worse since 2 years in the tech world is monstrously large.

 

But maybe that is just me and my opinion.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [WCCF] AMD’s New High Performance Processor Cores Coming Sometime in 2015 – Giving Up on Modular Architecture