Intel Broadwell Thread

Page 120 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Dufus

Senior member
Sep 20, 2010
675
119
101
Wonder if there's something weird with these chip's similar to the older Sandybridge needing PLL overvoltage to clock high.

Anyway hoping you get a good chip. Still unavailable where I live, looking like they might just be skipped instead for Skylake which are available.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Does anyone know of any OC rules that differ from Haswell? I don't know if there are enough of these out there for any guides to have been written. I'm going to be doing some reading tonight.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
So far I can't persuade this 5775C into any configuration that would exceed 65W. I've tried everything I can think of so far to no avail. Any setting that might result in the CPU exceeding 65W is either ignored or results in boot failure. It actually took me a while to figure out what was going on, but the damn thing was ignoring manual multiplier settings and throttling like crazy. I'm blaming poor BIOS support at this point, but prior to doing this swap BIOS F7 was installed on my GA-Z97X-UD5H-BK, so it's not for lack of preparation. The ultimate destination for this CPU was to be my wife's desktop, which is running a Haswell i3 and an HD 7750. The swap might come sooner rather than later if I can't persuade Gigabyte to fix the problem.

Edit: I forgot to mention that "Optimized Defaults" won't even let Windows 10 boot, that's how I have concluded that Broadwell support in this BIOS is probably half-baked. Very disappointing!
 
Last edited:

Dufus

Senior member
Sep 20, 2010
675
119
101
@crashtech if you run the attached and post / pm back the dump.txt file I'll take a look for you to see if there's any obvious register settings that are responsible.

Dumper.zip

How much vcore are you willing to use?
 
Last edited:

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
I'm not worried about short term overvolting, but right now turning up the voltage is counterproductive, it just causes more severe throttling.
 
Last edited:

Dufus

Senior member
Sep 20, 2010
675
119
101
Dumper.zip, forgot to add the link and by the time I edited the post you had replied. :/

Just asking about voltage in case of the need to write some software to set your voltage / ratio's and limits.

EDIT:
From your dump a 65W limit is set via power limiting memory mapped address.

RWEverything can be downloaded from here.
Portable 64-bit version is RwPortableX64V1.6.8.1.zip

Putting in wrong values can cause a BSOD so best to have everything saved before running.

Run the program and do the following to select the memory mapped addresses we are interested in.
1hxtoo.png


FED15900 is the usual address but can be different depending on 0x48 in the previous screenshot.
e1ezx2.png


These are the 2 locations we want.
m3uiq.png


Here's the PL1 setting which your dump shows as 65W with a time window of 0.977ms and clamping enabled.
vzzimq.png


So to change this to say 100W with a 28 second window for higher power, enabled and no clamping we would set


  • Power bits 14:0 = 100W x 8 = 800 = 0x320
  • Enable bit 15 = 1
  • Clamp enable bit 16 = 0
  • Time Window bits [23:22]*((2^[21:17])/1024) = 1.75*((2^14)/1024) then [23:22] = 11b, [21:17] = 01110b so 1101110b
This gives us 0x00DC8320 as the value for the register offset 0xA0 (....59A0).

To set 125W for PL2 with a small time Window for offset 0xA4 gives us 0x000683E8.

Notes:

  1. Bit 31 in offset 0xA4 acts as a lock bit for both 0xA0 and 0xA4 preventing further change until next boot.
  2. Clamp function permits throttling below the highest non turbo bin. With it disabled as recommended by Intel, only turbo bins will be throttled.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Edit: I forgot to mention that "Optimized Defaults" won't even let Windows 10 boot, that's how I have concluded that Broadwell support in this BIOS is probably half-baked. Very disappointing!

Agreed. Fugger reported that the best UEFI support for the i7-5775C came from an Asus board he tried when OCing his ES chip. He had better luck there than he did with the ASRock board he tried at first.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
For the moment I am asking Gigabyte for some help before I go mucking about manually changing register values. But if I was to make changes, I would want the power limit as far out of the way as possible, I am pretty sure my Corsair H110 AiO can dissipate well in excess of 150W at normal temps.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
For a minute there, I was going to ask how you were running a Broadwell-C chip on an H110 board. Then I realized you meant your AiO, not the motherboard . . .

There could be some confusion in the future between the corsair h110 lineup of AiO coolers and Intel H110 motherboards.
 

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
Late October 21.5'' iMac rumoured to include Broadwell-C/K:

Apple Planning to Launch 4K 21.5-Inch iMac in Late October

imacs.jpg


Apple will announce an updated 21.5-inch iMac with a 4K display at the end of October alongside the release of OS X El Capitan, reports 9to5Mac. The new iMac will then begin shipping out to customers in early November.

A new 21.5-inch Retina iMac has been anticipated for several months, following the discovery of code in OS X El Capitan that pointed towards a 21.5-inch machine with a resolution of 4096 x 2304. That code pointed towards Broadwell chips with Intel Iris Pro Graphics 6200 and AMD Radeon M380 - M395X discrete graphics, which may hint at the hardware we'll see in the updated iMac.

www.macrumors.com/2015/09/03/apple-4k-21-5-inch-imac-october-launch

Meanwhile ASUS opted for 35W Skylake-T + Maxwell dGPUs for their iMac 'Windows 10' rival.
 

Dufus

Senior member
Sep 20, 2010
675
119
101
You can change those limits to be higher or maybe try setting them to 0x00000000. I'm puzzled though why the 100W.exe didn't set those registers.

Yes, ultimately the BIOS should be fixed up. At least you can explain to them that MCHBAR +0x59A0 and +0x59A4 are not programmed correctly, remaining at 65W and 81W respectively.

Sounds like other issues too. Tweaktown forum has good Gigabyte support and user Stasio was the main guy to talk to, don't know if he is still there. He might know of a manufacturers beta BIOS that addresses the issues.

BIOS 8a
- eDRAM support for Broadwell (Z97)
- 17.Jun - 03.Jul 15

Download link on first page of linked post.
 
Last edited:

ehume

Golden Member
Nov 6, 2009
1,511
73
91
Sounds like other issues too. Tweaktown forum has good Gigabyte support and user Stasio was the main guy to talk to, don't know if he is still there. He might know of a manufacturers beta BIOS that addresses the issues.

Here.
 

Dufus

Senior member
Sep 20, 2010
675
119
101
Strange how those ES chips seem to overclock better than retail.

@crashtech, your multi control is set to use maximum turbo for anything higher than P-State 32x. Is this the control problem you were talking about?
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Not sure what you are asking, but as far as I can tell, any way the multiplier was set in the BIOS seemed to have no bearing on what Windows reported, whether it was via manual override or adjusting Turbo bins. Most of the time, an unfavorable setting would result in boot failure and a return to the BIOS to attempt a more favorable configuration.
 

Dufus

Senior member
Sep 20, 2010
675
119
101
Your dump settings show settings of 1-2 active cores at 39x, 3-4 active cores at 38x and for some reason 5-6 active cores at 36x. Any OS request for a multi higher than 32x will in this case result in a request for 39x or 38x depending on number of active cores.

Your cache multi is set for 36x only. IOW if your cores were running at 8x your cache would still be at 36x.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
I've pulled the 5775C from my system because it was crashing, not sure if it was the CPU or the mobo. At any rate, I can't perform any further tests, the data available will have to stand.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Bummer. Buyer beware until the motherboard OEMs can get better support for these chips. Some boards support them well and some don't.

Thanks for the data. You did it . . . FOR SCIENCE!!!!
 

zir_blazer

Golden Member
Jun 6, 2013
1,160
400
136
Broadwell BIOS support is the worst I saw in a while. For a ridiculous delayed launch, is hard to believe that they didn't had enough time to make sure it works properly.

So we can conclude by saying than Broadwell, the H97/Z97 Chipsets and SATA Express support were one big joke.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Broadwell BIOS support is the worst I saw in a while. For a ridiculous delayed launch, is hard to believe that they didn't had enough time to make sure it works properly.

So we can conclude by saying than Broadwell, the H97/Z97 Chipsets and SATA Express support were one big joke.

It just shows the good and bad mobo makers. Who got working support and who doesnt.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
I will be passing on Gigabyte this time around. The Z97X-UD5H-BK did manage to take my 4790K to 5.0GHz, so it's a good piece of hardware, but I can't say much for their firmware support.

Bummer. Buyer beware until the motherboard OEMs can get better support for these chips. Some boards support them well and some don't.

Thanks for the data. You did it . . . FOR SCIENCE!!!!

Thanks, lol!
 
Last edited:

ehume

Golden Member
Nov 6, 2009
1,511
73
91
It makes me wonder whose board I will get for Kaby Lake. I've always known Gigabyte BIOS was merely OK, and their software is second rate. But I thought their hardware was good. I had a taste of AS.S and decided that their board was heavy on the glitz. So who else make motherboards? MSI and ASRock? And is ASRock just AS.S Junior? It's a good thing I will have a year or more to decide.