Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [Videocardz] Leaked Zen slides - Identical performance to Broadwell-E
New Posts  All Forums:Forum Nav:

[Videocardz] Leaked Zen slides - Identical performance to Broadwell-E - Page 12

post #111 of 132
AMD saw the writing on the wall with the Bulldozer design. Couple that will low R&D and AMD just bit the bullet and started from scratch. How AMDs former executives even released the thing was odd.

That's why they didn't release any new desktop CPUs since 2012. Any improvements to the architecture just went into APUs since they couldn't abandon that area like desktops... that paid off with console design wins...

Once we get Zen APUs, APUs in general with be great.
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
post #112 of 132
Quote:
Originally Posted by davidelite10 View Post

Let's take a FX9590 and downclock it to 3.5ghz and compare it to a i7 4770k.

A perfectly reasonable thing to do if you are doing an IPC comparison.
Quote:
Originally Posted by davidelite10 View Post

But what they're showing is they can't meet intel, they have to lower that expectation to their level.

Zen is at least six months from launch and you cannot come to any sound conclusions of clock speed potential based on the clocks used for an IPC comparison between this engineering sample and a 6900K.
Quote:
Originally Posted by PontiacGTX View Post

1 Module on Bulldozer/VIshera/Streamroller fits SMT concept

CMT and SMT are completely different and two core CMT compared to a single core with two-way SMT have nothing in common other than they show up as the same number of logical cores to the OS.

A Bulldozer/Piledriver/Steamroller module is still two physical cores and a Zen or Intel core with SMT is still one.
Quote:
Originally Posted by Kuivamaa View Post

Both zen ES and the BW-E octocore are SMT designs with 8 cores that can be seen as 16 logical threads. If Zen trails BW in ST performance that would mean it makes up for it in superior SMT scaling. Frankly, this sounds improbable - Intel has been working on their SMT tech for more than 14 years.i doubt AMD can surpass them with their first attempt. It is more likely that zen has a slight ST perf advantage here and BW makes up for it with superior SMT scaling.

I strongly suspect that Zen does have somewhat superior SMT scaling, for two reasons:

- According to the block diagrams for both parts, Zen's execution core is wider, but has more fixed function units, which would rely more on SMT to gain full utilization. Intel's units are fewer, but much more multi-functional, which means some can be co-opted for increased ILP within a single thread. Zen, not being able to do this, would need more TLP to make up.

- A large barrier to Intel's SMT performance is cache contention. Zen has double the L1-I cache size and double the L2 cache size of contemporary Intel architectures.

I also suspect that Zen will prove to have slightly worse single threaded IPC (in Blender) and will be more limited than BW-E in well-threaded tasks with many dependencies or that otherwise requires strong inter-core bandwidth/low latency across more than four cores. The dual-module nature of Zen will introduce extra latency when one cluster of cores needs access to the other.
Quote:
Originally Posted by umeng2002 View Post

Yeah, just did some googleing:

"The L2 cache is actually split in two, one for each cluster, and communicating from each cluster to its own L2 cache takes 26 cycles. Communication between a cluster and the L2 cache attached to the other cluster is much slower, taking 190 cycles."

http://www.dualshockers.com/2014/03/11/naughty-dog-explains-ps4s-cpu-memory-and-more-in-detail-and-how-they-can-make-them-run-really-fast/

But the PS4 has NO L3 cache... so... what is the L3 latency for modern Intel CPUs?

I'm sure AMD has got that under control if they're planning 32 and 64 core versions... with as the article pointed out... with better branch prediction and making sure the right code is in the right L2 cache.

Modern CPUs tend to have L3 latencies between 30 and 60 cycles. Having to reach into another L3 to retrieve data will at least double that, probably more, but I don't think Zen will see anywhere near the penalty of the very different PS4 parts when going to a non-local LLC.
Quote:
Originally Posted by lolerk52 View Post

However, in the example you provided, they are both still two separate clusters, even if they are a different distance apart. while Intel's designs have them sit in a single cluster. That should be an inherent latency advantage.

Yes, and this is probably why a test like Blender was chosen. The clusters of cores could be on the otherside of the planet and that latency would not matter because rendering like this is perfectly parallel...the cores barely need to communicate with each other at all to get the task done.
Quote:
Originally Posted by Ha-Nocri View Post

So, what's the blender benchmark about? Can we estimate gaming performance based on that, or it measures something else?

Blender is a rendering application. It's platform agnostic and almost perfectly parallelizable.

Probably not a good representation of gaming or other non embarrassingly parallel workloads.
Quote:
Originally Posted by lolerk52 View Post

Blender is quite heavy on FP performance IIRC, which games are absolutely not.

This too.
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 #113 of 132
Good points on the possible effects of inter-core cache latency for things other than parallel tasks.
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
post #114 of 132
Quote:
Originally Posted by Blameless View Post



I also suspect that Zen will prove to have slightly worse single threaded IPC (in Blender) and will be more limited than BW-E in well-threaded tasks with many dependencies or that otherwise requires strong inter-core bandwidth/low latency across more than four cores. The dual-module nature of Zen will introduce extra latency when one cluster of cores needs access to the other.
Modern CPUs tend to have L3 latencies between 30 and 60 cycles. Having to reach into another L3 to retrieve data will at least double that, probably more, but I don't think Zen will see anywhere near the penalty of the very different PS4 parts when going to a non-local LLC.

Thing is that features like large L2 cache (which Zen reportedly has in bigger quantities than BW and SL) also benefit IPC in many ways and for many workloads. Obviously this new processor does not beat BW-E clock for clock in both ST and MT perf. Given its width and the fact its L3 is not totally unified (for the 8 core part that is) I tend to believe that at least for blender, it performs on a very similar level to BW both in ST and MT. We'll see.
Mastodon Ryzen
(12 items)
 
HP Z220
(8 items)
 
 
CPUMotherboardGraphicsRAM
R7 1800X Asus Crosshair VI Hero Sapphire RX Vega 64 reference Gskill TridentZ 
Hard DriveOptical DriveCoolingOS
Pny SSD 240GB Crucial MX100 CM Nepton 280L Win 10 
MonitorPowerCaseMouse
Acer Predator XG270HU Freesync XFX 750W Pro HAF XM Logitech G502 
CPUMotherboardGraphicsCooling
i7 3770 HP Quadro K2000 HP 
OSPowerCaseMouse
Win 7  HP 400W HP CMT RAT 7 
  hide details  
Reply
Mastodon Ryzen
(12 items)
 
HP Z220
(8 items)
 
 
CPUMotherboardGraphicsRAM
R7 1800X Asus Crosshair VI Hero Sapphire RX Vega 64 reference Gskill TridentZ 
Hard DriveOptical DriveCoolingOS
Pny SSD 240GB Crucial MX100 CM Nepton 280L Win 10 
MonitorPowerCaseMouse
Acer Predator XG270HU Freesync XFX 750W Pro HAF XM Logitech G502 
CPUMotherboardGraphicsCooling
i7 3770 HP Quadro K2000 HP 
OSPowerCaseMouse
Win 7  HP 400W HP CMT RAT 7 
  hide details  
Reply
post #115 of 132
There is always a catch with AMD products. I expect Zen will have a major flow. Won't be able to OC, or it's power consumption... or something....
post #116 of 132
Quote:
Originally Posted by Ha-Nocri View Post

There is always a catch with AMD products. I expect Zen will have a major flow. Won't be able to OC, or it's power consumption... or something....

Based on this demo and AMD's claims, I think the biggest concern would be the quality of the silicon... The issue with Bulldozer+ was, fundamentally, the design.
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
post #117 of 132
Thread Starter 
Quote:
Originally Posted by umeng2002 View Post

Based on this demo and AMD's claims, I think the biggest concern would be the quality of the silicon... The issue with Bulldozer+ was, fundamentally, the design.
The hell is quality of silicon supposed to mean?
post #118 of 132
Quote:
Originally Posted by lolerk52 View Post

The hell is quality of silicon supposed to mean?
alot of bad batches?
old dog
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 2600K@4.60@1.35v P8Z68-V PRO EVGA 1080 Classified DT Corsair 8GB 1600 
Hard DriveHard DriveHard DriveOptical Drive
3.5Tb raid 0 120 GB SSD 1 TB HDD the internet 
CoolingOSMonitorKeyboard
Custom loop Windows 10 Asus 144hz Ducky Shine 6 Snow White Edition 
PowerCaseMouseAudio
Azza titan 1000w nzxt switch 810 mmo 7 sound blaster z 
  hide details  
Reply
old dog
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 2600K@4.60@1.35v P8Z68-V PRO EVGA 1080 Classified DT Corsair 8GB 1600 
Hard DriveHard DriveHard DriveOptical Drive
3.5Tb raid 0 120 GB SSD 1 TB HDD the internet 
CoolingOSMonitorKeyboard
Custom loop Windows 10 Asus 144hz Ducky Shine 6 Snow White Edition 
PowerCaseMouseAudio
Azza titan 1000w nzxt switch 810 mmo 7 sound blaster z 
  hide details  
Reply
post #119 of 132
The only thing I am worried is that all the info even the official press conference said "identical clock speeds" and the i7 was brought down to 3.2ghz I think, which would mean AMD are still not able to get to the clock speeds to Intel, which in reality may mean that the performance is a lot worse in games, hope that's not the case, but every source there is "identical clock speeds" and they always around 3.2ghz. And Kaby lake will be doing 4.2ghz, which is no small difference. if they can get to 3.8-3.9 maybe won't be a difference. I sure hope they succeed because without any competition there is no point of pushing forward for Intel.
post #120 of 132
Quote:
Originally Posted by lolerk52 View Post

The hell is quality of silicon supposed to mean?

Junk leakage, bad yields, etc.
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  hide details  
Reply
AMD Box
(15 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8320E @ 4.6 GHz +0.356250v offset Asus Sabertooth 990FX Rev1 eVGA GTX 970 SC ACX2.0 Patriot Viper Xtreme 2x4 GB 1600LL 8-9-8-24 1T 
Hard DriveHard DriveHard DriveCooling
Samsung 840 EVO WD Black 1 TB 32MB cache FALS WD Blue 1 TB 7200rpm EZEX Corsair H80i 
OSMonitorKeyboardPower
Windows 10 x64 HP LP2475w Logitech Illuminated Corsair TX750  
CaseMouseAudio
You don't want to know Logitech G9x Creative Sound Blaster Z 
  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 › [Videocardz] Leaked Zen slides - Identical performance to Broadwell-E