Why did Intel even release Broadwell-E?

Toggle sidebar Toggle sidebar
P

Potato Joe

Reputable
Jun 3, 2015
151
0
4,690
The IPC gains over Haswell are negligible and the additional voltage needed to hit the same clock frequencies ruin any sort of energy efficiency gain. From what I've seen, a lot of people with 6800K need 1.35V minimum to hit 4.4GHz, and that's the low-end, meanwhile a good number of 5820K get by at 1.27V or less for the same frequency.

What was the point? And more importantly, why are people upgrading from Haswell to Broadwell? They're basically the same except for the heat spreader design.

Sidenote: I accidentally typed "upgrading from Broadwell to Haswell" the first time I went to post this. Caught it in preview.
 
Sort by date Sort by votes
P

Potato Joe

Reputable
Jun 3, 2015
151
0
4,690


Sorry for not specifying that I actually meant Haswell-E, although the Z97 compatible Broadwell chips are garbage as well compared to the equivalent Haswell.

tl;dr
Broadwell is literally not an improvement over Haswell. Why, as a consumer, should anyone consider Broadwell, in either standard desktop or enthusiast class? Skylake has it beat in the non-enthusiast class, and Haswell has a much better price-performance ratio on the X99 platform.
 
Upvote 0 Downvote
k1114

k1114

Titan
May 2, 2010
14,240
93
75,540
I understood it just fine. But the mainstream bw was a mostly ignored released and intel knew it would happen. There's a reason there was hardly any mainstream bw cpus released. There r&d results were just small and they moved on.

Hedt is different though as it's really dependent on the higher end xeon segment which it's released alongside. It's the reason why hedt is behind in generation vs mainstream because that platform has a need for longevity. But also at the same time, regular releases is part of the cpu industry which is the reason for the haswell refresh as well has the skylake refresh, aka kaby lake, and the change to a tick-tock-ding cycle as I call it.

However, it was still an ipc increase and improvement in power consumption at stock speeds. It's kinda like sb to ib when oc is involved and puts them on the same level (as well as that comparison also being a tick). Although the price increase makes it less appealing and was a bad move imo. It makes hw-e a more worthy option.

Why people upgrade from hw-e to bw-e? You probably should ask them. I don't see anyone here suggesting the sidegrade unless they have no idea what they are talking about.
 
Upvote 0 Downvote
Ecky

Ecky

Illustrious
Feb 3, 2013
7,592
2
38,960
You really shouldn't compare voltage between Haswell and Broadwell, because they're built on completely different processes.

Broadwell is more energy efficient (significantly so at lower frequencies) because it's built on Intel's 14nm process. The efficiency gap narrows at higher frequencies but Intel was able to stuff more cores into the same power envelope in servers with Broadwell, which is a Big Deal, and in laptops it significantly improved battery life. On socket 1150, the decreased die size allowed Intel to pack a humongous iGPU onto the chip, which would not have fit with Haswell / 22nm.

For overclocking? Yeah, no real benefit because Intel tuned the process for efficiency and not raw clockspeed.
.
 
Upvote 1 Downvote
P

Potato Joe

Reputable
Jun 3, 2015
151
0
4,690


The reason I bring up voltage is because a higher voltage at the same amp rating means Broadwell can use potentially more power at the same clock speed if you get a less than average chip. Not even a bad one, just less than average.

Also, a higher voltage will kill Broadwell faster than Haswell because, since Haswell is built on a larger manufacturing process, it's more resistant to electromigration. Broadwell for overclocking is a terrible idea in general.
 
Upvote 0 Downvote
P

Potato Joe

Reputable
Jun 3, 2015
151
0
4,690


Actually if I recall correctly there was something straight from Intel about the max voltage for 22nm to run at without accelerating electromigration being 1.3 and 14nm being 1.25. If I'm wrong, I'm wrong, but I'm about 90% certain that info came straight from Intel.

After researching further, yes, this has been confirmed. on 22nm 1.3V is as high as you can go without shortening the lifespan of your CPU, while at 14nm 1.25V is as high as you can go without shortening your CPU's useful life, assuming you're running the overclock 24/7.

So yeah... Even if we ignore the differences in characteristics, straight from Intel, smaller manufacturing processes are less tolerant to overvoltage.

Also, what are you talking about for amp ratings? Amps are amps, regardless of what they're running through.
 
Upvote 0 Downvote
P

Potato Joe

Reputable
Jun 3, 2015
151
0
4,690
...Huh. I can't seem to find it.

With that said, why does Skylake run as high as 1.4v stock? This is confusing to me seeing as Haswell runs (depending on the specific CPU) 1v flat to 1.2v
 
Upvote 0 Downvote
Ecky

Ecky

Illustrious
Feb 3, 2013
7,592
2
38,960
Different architecture, different process. Skylake draws around the same amount of power at 1.4v as Haswell does running at ~1.20-1.25v at the same clockspeed, and since Intel warranties them both for the same time, it's probably safe to conclude that the reliability of both will be similar.

It's also important to note that Haswell has an intergrated voltage regulator, while Skylake doesn't, so "1.4v" and "1.25v" are probably not even talking about the same thing.
 
Upvote 0 Downvote
P

Potato Joe

Reputable
Jun 3, 2015
151
0
4,690
It was in response to a question on this forum. I'll find it eventually, probably after it no longer matters.

And 1.4v will always be 1.4v regardless of what it's running through as long as the regulator is working correctly.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom