Buy now or buy later? Haswell sleep bug

[goofy0115]

Hi everyone!

 

I'm about to build my new pc and I was originally considering i5-4670K with a z87 motherboard.

However I recently found out about the s3 sleep bug. Is this sleep bug a really big deal?

Should I still buy Haswell cpu or wait till they have the c2 stepping? (When will the C2 version be available to us? I know they're sending them out July 15th or something)

 

Or

 

Should I just buy i7-3770K with z77 motherboard instead?

 

In terms of overclocking difficulty level, is it any different for the 3rd generation intel processor and 4th?

 

Sorry about all these questions.

 

Thanks so much!

[Coombzy]

overclocking wise haswell does run hotter but if you don't care about that I would go for it although you are comparing getting an i5 to an i7 so idk perhaps think of the 4770K instead can't be that much more.

[Glenwing]

The bug isn't that big of a deal, although it depends on what you might do with your computer.  It is just that certain storage devices (not all) when plugged into a native USB 3.0 port and the computer is put to sleep, the devices has to be re-seated.  Not a big deal honestly and you probably won't even run into it.

[goofy0115]

When you mention native usb 3.0 ports, what does that mean? Like the ones in the back of the computer which are directly connected to the motherboard?

Also "devices have to be reseated", does that mean just unplugging and plugging it back in?

 

Also, does every haswell cpu have this bug? or is it like 1/5 or something like that?

 

Thanks so much for the replies

[DeathAndPain]

When you mention native usb 3.0 ports, what does that mean?

"Native USB 3.0 ports" means USB 3.0 ports that are provided by the Intel chipset. There are also mainboards out there that have additional third-party USB 3.0 chipsets soldered upon them (e.g. to increase the number of USB 3.0 ports of the mainboard beyond the number the Intel chipset offers). Ports driven by such third party chips are not affected for obvious reasons.

Also "devices have to be reseated", does that mean just unplugging and plugging it back in?

That is probably what Glenwing meant. However, it is not even true.

First of all, only USB 3.0 sticks that use a certain-brand USB 3.0 controller chip are affected. Any other USB 3.0 devices, including USB 3.0 sticks that use a different controller chip, are unaffected.

Second, there is no need to re-seat the stick even if you are using such a stick. The only thing that happens is that when the computer wakes up from a certain sleep state, it re-enumerates the stick again. This means that it acts as if the stick had been removed and instantly plugged back in, having it vanish in Windows Explorer for a second or so before it re-appears. Normally this has no consequences whatsoever; you will not even notice it. However, if you are running a program that tries to access the stick instantly after your computer waking back up, then these programs will stumble over the fact that the stick is gone for a second. Most applications, however, do not access the stick again this quickly after wakeup. Iirc this even goes for Microsoft Word if you have a document opened that is located on the stick.

Also, does every haswell cpu have this bug? or is it like 1/5 or something like that?

No Haswell CPU has this bug. The bug is located in the corresponding chipset that is soldered upon the mainboard and also originates from Intel. There are no third-party chipsets for Haswell CPUs, and there will likely never be any, so all mainboard manufacturers have to use Intel chipsets for their Haswell mainboards. Intel offers a selection of different Haswell chipsets (with differing features), but all of them share the bug.

Since the bug is a design flaw, every chipset has it. There is no such thing like "having a chance" of a chipset to have it.

Oh, and concerning what Coombzy said: Haswell does not run hotter. Neither when overclocking nor normally. This is another widespread misconception.

Haswell does have a higher maximum power consumption, but only when its AVX units are under full load. AVX is the new instruction set that its predecessor (Ivy Bridge) does not have in the first place. However, practically no existing software uses AVX yet, so with these, Haswell does not run hotter. Future software may use AVX, but the gain in performance from AVX-optimized code is approximately a whopping 70%, while the increase in heat and power consumption is way lower. So the performance-to-power-ratio is better for Haswell in all cases! In those cases in which it runs hotter than an Ivy Bridge it does so because it delivers an amount of performance Ivy Bridge could only dream about.

Meanwhile, Haswell also has additional new power states that cause it to consume even less power than Ivy Bridge in idle state.

There is really no reason to go for the obsolete Ivy Bridge platform anymore.

[Exotath]

My 1050T system has this problem in Win 8....can't figure it out for the life of me. So...I turned off sleep.

[Torg73]

My 1050T system has this problem in Win 8....can't figure it out for the life of me. So...I turned off sleep.

I turned off the sleep too when I got my new build up and running I switched it back on, works ok now.  Some of the sleep problem could be your power supply.  When picking a new PSU look to see if it compatible with Haswell, there's list on the net, just Google it.

[DeathAndPain]

The power supply thing also exists, but that is a totally different issue. It is related to the new low-power-states that Haswell supports and that I mentioned. This is not a bug, it is a feature. Haswell can reduce its power consumption on certain power rail below the minimum expectancy of the PSU. This can lead to crashes.

However, few halfway modern PSUs encounter this problem, even if they are not yet Haswell-certified and way older than Haswell itself, because for reasons of lowering production costs, most PSUs use only one shared rail. In these cases, no problems will occur.

If you are using an old PSU that has this problem, there is still no need to disable sleep altogether. It suffices to disable the new sleep states in the BIOS/UEFI of the mainboard. Don't ask how the corresponding setting is named, as this obviously depends on how the BIOS/UEFI programmer opted to name it and can vary from mainboard to mainboard.

[warmmilk]

with ssd's these days, i never put my comp to sleep... I just shut it down.  i can wait 15ish sec's from pushing the power button to using my comp