Devil’s Canyon overclocking? Yay or nay?

Devil’s Canyon is the code name for the two CPUs from Intel aimed at enthusiast overclocking. They are still based on the same Haswell micro architecture though. What’s different is that they have better TIM – something Intel calls next generation polymer based TIM – and few more capacitors underneath the CPU to smooth out the power delivery to the CPU. Would just these two changes be enough to win back the hearts of the enthusiasts.

However, in a previous post I mentioned that the Devil’s Canyon CPUs – especially the 4790K - weren’t performing up to my expectations. That was completely based on a review posted on Hexus, which didn’t even explore the overclocking capabilities of the CPU. It seemed like a rushed up job and they just wanted the initial swarm of people looking for the reviews to come to their website.

But today, two more reviews went online; one from PCPerspective and the other one from Digital Storm. Just like one would expect, they do explore the overclocking capabilities of the CPU. But neither sample was able to make me feel enthusiastic about the CPU. The one PCPerspective got only managed to do 4.7GHz at 1.365V. The one Digital Storm got only managed to do 4.8GHz at 1.370V. Those are pretty high voltages for CPUs built on the 22nm fabrication process. And neither managed to hit the magic 5GHz. In fact, there are people who own 4770K’s that can do 5GHz at as little as 1,3V but that is the cream of the crop CPUs. However, despite the high voltages, the coolers they used – AIO water cooler by PCPerspective and a custom water loop by Digital Storm – seem to have managed to tame the temps down to reasonable levels. At least that’s something. The new TIM seems to do it’s job just fine. Still, you would probably get much better temps with a delid and Coollaboratory Liquid Ultra TIM.

Since these CPUs come with a guaranteed 4.4GHz clock speed – at least in the case of the 4790K, every 4790K will be able to overclock past my current CPU – which can do a measly 4.3GHz – while keeping the temps lower. That means lower fan noise as well. If you look from that perspective, it seems upgrading to the 4790K still has some merits, even if I get a dud chip that can merely do 4.6GHz at 1.3V. The only way to tame the temperature with my current CPU is to delid, which I am hesitant to do as it would make reselling harder – especially with the clumsy clock scaling of my chip.

Even though the initial impressions are not that positive, I would still be waiting to see how the retail chips would overclock and the pricing of the chips in Japan. One would probably argue that Haswell-E is the way to go, but that could require a substantial amount of money.

[Gadget] Bought A-Data XPG V2 2400MHz CL11 2x8GB kit

In the previous post I wrote how I finally managed to sell the old Corsair Vengeance RAM. I sold them to a total of JPY12,000. Before sending the sticks to the buyer, I had to buy new sticks for my rig.

I've written this before as well - RAM prices have skyrocketed in the past few months. So my options were pretty small. I originally wanted to get the ones with the shortest heatsinks. The shortest kit available was the Crucial Ballistix Tactical LP 1600MHz one, but the prices of it had almost doubled in the last few months and now were selling for a whopping JPY25,000 or more. They used to be around JPY14,000 few months back - ironically the cheapest kit back then. Current prices made it a no buy for me.

The other regular height RAMs were all around JPY20,000 or more, unless I went for the 1600MHz CL10 sticks which were available for around JPY16,000. Of course I didn't want to go for slower RAM than what I had before. So those options had to be cut from the list.

There were some A-DATA XPG V1 1600MHz CL9 ones for JPY16,000 or so. And then the XPG V1 2133 ones for JPY19,000 or so. But my eyes got locked to the 2400MHz CL11 XPG V2 sticks which were going for JPY18,817 at Amazon. Only thing was that they were a bit too tall. Not as tall as the Vengeance sticks that I had before but still, taller than  regular DIMMs. When I read the Kitguru review (for 2800MHz version of these sticks), I found out that the heatsink can be removed but they didn't go into detail on how to do that. But the possibility gave me hope. Not sure if I would really go so far. Removal of heatsinks will definitely void warranty, and these sticks have lifetime warranty.

Disgusted by the prices of the low profile RAM, I ordered the A-Data kit from Amazon and it arrived the next day. (Funny thing is that ever since I bought these sticks, the prices at Amazon has gone up by JPY1,500!!!)

[Rant] Haswell refresh?

We all know how much of a crappy chip Haswell is when it comes to overclocking. (Refer this post about temps and this post about FIVR). All I have managed is a measly 4.3GHz out of my Core i7 4770K. (Refer this post) My old 2600K was capable of 4.5GHz, so the Haswell upgrade hasn’t given me a big boost. It’s still faster but not fast enough.

Intel is releasing Haswell Refresh CPUs soon and already MSI and Asus have claimed that their current boards will support these new CPUs with a bios update. Why this is important is because it seems Intel is going to release a new chipset – the 9 series – along with these new CPUs. That doesn’t sound right though. The 8-series is very strong, and there isn’t anything architecturally new with Haswell Refresh which requires a new chipset. They should do it for Broadwell. This is a weird move if they release 9 series this soon.

Anyways, what's actually new about these CPUs? Are the faster? Are they going to overclock higher? Will the die be soldered to the heat spreader?

From what I can see, they are only bringing a slight clock boost to the existing line up and Iris Pro graphics (perhaps).

[Rant] Got a BSOD in months!!!

A few months back, BSODs were a normal thing with my new rig. That was because I was trying to overclock my Haswell Core i7 4770K CPU. I spent a whole month trying to get it to 4.4GHz and finally I gave up and moved on. Now I'm a happily sailing away with a Clock speed of 4.3GHz.

A few days ago I noticed that there was a new BIOS (1002) for the motherboard (Asus Maximus VI Hero) and last night I wanted to try it out. Actually, I skipped the previous one as well. I wasn't going to see if it would overclock better than it used to. I just wanted to update for the sake of updating it.

First of all, I took screenshots of all the BIOS pages because updating would erase all of that. That's a dumb move on Asus's end, right? You can save it to a USB drive as a profile but sometimes when they version the BIOS up, the profiles won't be recognized anymore. Again, quite a dumb thing.

Anyways, just to be safe, I took photos of the bios pages. I could have saved them on to USB drive as screenshots using the shortcut key but that would mean I would have to use the tablet to view all those images. I didn't feel like using the tablet at that time so I just took photos from the phone.

Then it was the time to update the bios. I copied the bios file to the USB flash drive and went into the BIOS. I used the EZ Flash utility inside the BIOS to open the bios file and it took over the rest of it.

After flashing the new bios, it rebooted the PC. Before successfully rebooting, it had to go through 5 POSTs. The PC powers up, the after a couple do seconds it powers down - and this went on for 5 cycles. Man, I hate when that happens. You don't know if something went wrong or not. Anyways, it finally managed to POST and then another screen showed up saying that it is updating iROG... whatever nonsense that means. It took about a minute to complete.

After that it rebooted automatically and I went into the bios straight away and changed the settings back to where they were before. Everything seemed fine so I rebooted to Windows. Except, it didn't work as I expected. It gave me a BSOD just before trying to log into the desktop. Weird, I thought.

[Rant] I'm not going to delid my 4770K

Why? Because there is no point. My chip is voltage limited. I cannot get 4.4GHz stable with below 1.3V and I don't want to feed it more volts. Sure some people don't care but I do. I don't want the power consumption of the CPU to increase by 50% just by getting that last 100MHz stable. Sure, if you run the CPU cooler, you can reduce the power consumption. In that sense it is worth delidding because that would allow you to du  the CPU cooler at the same settings. Plus, I have the Liquid Ultra thermal paste with me.

But, no.

I'm going to do it differently. I'm going to see what Intel is going to with Haswell Refresh that's coming in next year. I'm sure it wont be much faster than what's available right now. But it might clock better. Again, there is good chance that I would lose the silicon lottery once more. That's become a bad habit of mine. But I will no do anything else with this CPU. I wont try to tweak it anymore. I wont buy faster RAM because it would not be noticeable. I won't add another CPU fan to make it run cooler. All those are diminishing returns. I would put all best I have towards the G710+ keyboard, G602 wireless mouse and a R9-290X graphics card.

Oh btw, I ordered a new set of headphones. It w's neither the AD700X nor the G430. It's the AD500X. I felt that I was not audiophile enough to notice the difference between the 700X and the 500X. the 500X is about half the price of the 700X. These e are Japanese prices, so usually much less than rest of the world. I'll right about them on a separate post.

[Rant] Should I upgrade the RAM on my Haswell rig?

When I upgraded to Haswell, I decided to keep using the RAM from the old rig. They are two kits of 4GBx2 Corsair Vengeance 1600MHz RAM. I tried overclocking those RAM, but they don’t want to deviate from the stock settings. Funny thing is though, they can work fine at XMP settings with just 1.35V, down from the 1.50V.

With the previous couple of generations, RAM speed didn’t make any difference in performance. But the integrated memory controller (IMC) on Haswell CPUs is strong and can run RAM even at 3GHz. (Sure it can cripple the core overclock of the CPU). Anandtech published an article saying that the sweet spot is between 1866MHz and 2400MHz.

Even then, the performance advantage is negligible. But for me, I can upgrade RAM and improve some other aspects of the rig.

• First is, by buying a 8GBx2 kit, I can lower the power consumption as well as the stress on the IMC. Usually, when you populate all the memory slots, it stresses the IMC more.
• Second is, if I can go with low profile modules, I can mount a 3rd fan on the CPU cooler for improved cooling performance. But it won’t make a big difference.

So which kit fits all these criteria? I could find only one such kit. Crucial Ballistix Tactical LP kit. They are low profile (not just low profile heat sink, but the PCB itself it low profile; even with the heatsink they are shorter than standard RAM with no heatsink), cheaper than the kits in the same league (JPY14,470), comes with life-time warranty and uses only 1.35V. They operate at 1600MHz 8-8-8-24 timings. The speed is not great, but reviews say that if you push the volts to 1.5V, you can easily clock them to 2133MHz, which is very good considering they still cost less than 2133MHz sticks. However, I’ve been bitten by putting my faith in these reviewers, who usually get cherry picked samples. A lot of people managed to run the Vengeance sticks at 2000MHz or above but mine wouldn’t budge.

On the second hand, the 2400MHz sticks are well over JPY 20,000 and come with tall heat sinks. I won’t be able to install a 3rd fan in there.

But I still haven’t decided whether to get these or not. The RAM prices have gone up in the last few days and perhaps they won’t come down anytime soon. The prices of these Crucial sticks have not gone up. Maybe I should order them now, and sell my current sticks later when the prices are up.

[Article] Haswell, VDroop and Load Line Calibration

When Intel CPUs are stressed with heavy load, the Core voltage (Vcore) drooped slightly to relieve the voltage controller on the motherboard. We call this VDroop. It is not a sudden drop, but it is maintained throughout the period the CPU is stressed, hence is was not called a VDrop.

The amount of VDroop depended on how much Voltage the CPU was using up. Thus, when the CPU was overclocked - meaning, the Vcore was higher than nominal figure, you could see a bigger VDroop. While this was according to Intel's design specification, it was not a good thing for overclockers.

The CPU becomes unstable when the Vcore drops, especially when it is under load. Besides, all these overclocking boards had way better voltage regulating circuitry and the sudden increase in power draw from the CPU was nothing they couldn't handle. Thus, motherboard manufacturers did something to counter the VDroop. The common term they use is Load Line Calibration. I'm not sure how it works exactly, but it seems to add up more volts when the Vcore droops. Usually we can decide how much voltage the motherboard should counter apply by selecting the level of LLC, so it is not really "intelligent". Further, I'm sure the other voltages drooped a bit as well, but not in significant amounts. It was the Vcore that everyone was concerned about.

1st of June, 2013 arrived and came Haswell. Haswell brought a lot of new things to the table. Voltage regulation was one of that. Instead of depending on the motherboard's voltage regulator to split the +12V into individual component voltages such as Vcore, Vmem, VCCIO, VCCSA, Uncore voltage etc and input them separately, Haswell asked for only one specific voltage from the motherboard. It's called the VRIN or VCCIN or Input Voltage. (It's about time they all agreed on a common term!). In Haswell, there is a voltage regulator inside the CPU (called FIVR) and that's what takes care of splitting this input voltage to different individual component voltages. They have done this to improve the idle power consumption. Dropping 12V down to 0.xxxV is difficult for any voltage regulator. But it is much easier when you drop it from about 1.8V. The default VCCIN is 1.8V. 

VDroop is still there with Haswell. But it doesn't apply to Vcore like previous generations. It applies to the Input Voltage - the only voltage inputted to the CPU from by the motherboard. The default Input Voltage is high enough so that for a casual overclock, you don't have to worry about VDroop. But at high overclocks (which means, many component voltages will be running above stock voltages) you will run into issues. I think, you will see two phenomena caused by lack of Input Voltage.

• First one is, that the FIVR will not have enough juice to do its magic. This will cause a dip in any of the volts that would result in instability of the CPU. You will almost always get a BSOD with 0x124 stop code. This is probably why most people say that increasing the Input Voltage allowed them to use a less Vcore to stabilize the CPU.

• Second phenomenon is that the FIVR will completely shutdown. This will cause the PC to reboot out of nowhere. You won't see a BSOD or any warning. It will be as if the power went off and came back on.

I've seen some reviewers say that you don't see VDroop on certain motherboards even though there is no LLC setting. They say by looking at the Vcore droop. But Vcore doesn't droop with Haswell. They are looking at the wrong voltage. You have to monitor the Input Voltage. Then you will see that LLC still is required. But since it doesn't affect the Vcore directly, if the Input Voltage is high enough, you will be able to get away without fiddling with the LLC.

[Article] The ultimate stress test for Haswell CPUs - Prime 95 with AVX2/FMA3 support

Overclocking Haswell CPUs is a lot of work. Checking whether the overclock is stable or not is very time consuming. Some people go for a couple of hours of stress testing because most of the times that is enough to ensure that the CPU will run stably in every real world situation. But there are some people who cannot sleep if they know that one of their favorite stress testing apps didn't pass.

When it comes to stress testing applications, we are hearing a whole new story with Haswell. "This stress testing application is not certified for Haswell." Basically, that's what people say when they hear that you are using Prime 95 or LinX or OCCT. If by that they mean is that these applications don't  test every part of the CPU, the fine. But ironically, these apps can crash the CPU way quicker than those Certified applications. It makes one wonder, if Intel's CPUs don't work as they should. I mean, why cannot you simply run anything? Applications don't have to be certified for a CPU.

If it was the former, that is the application doesn't stress all the parts of the CPU, then the developers of Prime95 is finally fixing that. The current stable version, 27.9, supports up to AVX instructions. But no AVX 2 or FMA3 instructions. The developers have released an evaluation version of the upcoming version 28 which supports all these new instructions in Haswell. You can download it from THIS LINK.

Result?

Simply don't run the Small FFT test! On most setups, if you are above 1.20V, you'll be at instant throttling state. With 4.3GHz/1.24V, it would go to 97C within a couple of seconds. Well, I have my CPU fans at 1000RPM instead of the max 2400RPM, so it takes a couple of seconds to ramp up the fan speed. Maybe if I started at full fan speed, it might not hit that high. But still it will definitely be in the 90's,

But I don't stress test using Small FFT. I've found that almost always it is the in-place large FFT test that fails with my CPU. I ran it for a couple of hours, with each test running for only 3 minutes. The temps weren't that bad, however yesterday's weather was a tad cooler. 

Note: The Vcore written in the notepad is wrong. It should read 1.250V (1.264V in CPU-Z).

If you are running this, just be aware of the temps. There is no other stress test that can heat up the Haswell CPUs this much.

[Article] Asus’ misleading marketing about ROG Z87 boards and OC Tuner

Everyone on the Internet must already be knowing that I've upgraded to Haswell because I've written a ton of articles. When I was deciding on which motherboard to get, the capabilities of the AISuite software that comes bundled with the Asus motherboard played a big part. JJ of Asus marketing had posted a video showing the auto-overclocking features of the software some time back and I referred it before buying my Maximus VI Hero motherboard. Everything looked nice and functional. The auto-overclocking feature would test the maximum CPU clock speed you can achieve by running a stress test with every step of increase in multiplier. It would even try to overclock depending on the number of cores used. Please watch the video for all the details. It looks fascinating.

1-Click Overclock to 4.8Ghz - 4-Way Optimization on ASUS Z87

But after I bought the motherboard and decided to try it out, it wasn’t available in the AISuite III software that comes for this particular motherboard. It only had 3 predetermined overclocking profiles called CPU Level Up; not the intelligent auto-overclocking feature (OC Tuner) described in the above video. (This page explains the 3 CPU Level Up values).

When I asked around, the answer I got was that the target audience of ROG boards is different and ROG motherboards don’t support it. That’s total BS. I mean, ROG is their high-end motherboard series, but they would put a crippled software and charge more because the audience of the motherboard is more “geeky”? You are paying more to get less? Looks to me their target audience is “idiots”.

I’ve seen many people complain about this in the forums and the comments section of the above YouTube video. I hope Asus will release an updated AISuite/BIOS to bring that functionality to the ROG motherboards. If they don’t, I’m quite disappointed in Asus. It’s not that it is the end of the world. You have to use the BIOS anyways. Those auto-overclocking settings are almost always not stable when stressed out. But at least you get a starting point to build your overclock on top of. I paid for that feature; I want that feature. Besides, Asus doesn’t state anywhere that the ROG boards don’t come with that feature. JJ doesn’t say that in his video.

[Article] Not AVX Stable @ 4.4GHz, are we now? Shame!

I was able to run Intel XTU at 4.4GHz for 12Hrs+ and I thought it was stable. How wrong I was! It crashed in Handbrake after encoding videos for many hours. Once after 10hrs! After that I figured that there was something that I must have overlooked. When I asked around, I came to understand that XTU doesn’t use AVX in its stress test. Handbrake does, but still it doesn’t bombard it with AVX instructions like Prime95 does. BTW, Prime 95 27.9 (which is the latest version out with AVX support) crashes easily with the settings that XTU ran for hours and hours without failing.

Then I figured that it could be stable everywhere but AVX. One way to check it was to run Prime95 without AVX. That’s version 26.6. I ran it for hours and it didn’t fail. Now I was sure that it was the AVX instructions that caused the issue. 1.285V just isn’t enough when the components that makes AVX instructions work also become alive inside the CPU. I’m sure I would probably be able to escape by using Adaptive Vcore which automatically increases the Vcore by 0.1V when AVX instructions are heavily used. I tried 1.285V Adaptive, but it still failed in Handbrake. Probably the voltage fluctuation is also not a good thing.

Since then I’ve come to accept that you cannot have everything you want in life and 100MHz isn’t worth that increase in Vcore and temps (and increase in fan speed). So, now I’m going to settle with my 4.3GHz OC and tweak it a little bit to get every bit of power I can get with it. I’ve successfully managed to up the Uncore to 39x at just 1.08V and Memory to 1600MHz @ XMP. I’m also in the process of tweaking the Digi+ power delivery settings (image below) to lower the power consumption and the temps as much as possible without crippling stability. The following screenshot was taken while running Handbrake. Look at the fan RPM – bottom right hand side. It’s completely silent. That’s just great. I won’t be able to achieve that with close to 1.3V. The temps it shows is not very accurate though. Those are not the core temps. Core temps hover between 60C and 70C.

I haven’t done a lot of testing, but after dropping the LLC to level 6, enabling Active Frequency Mode and setting CPU Power Phase Control to optimized, at least it is stable in Intel Burn Test, Handbrake and Crysis 3. If it passes 6hrs of Prime95, I say I’m stable enough.

[Article] Haswell stress testing using Intel Extreme Tuning Utility(XTU)

I'm still trying to find the sweet spot for my Haswell CPU overclock. Prime95 has been failing on my in a weird way. Sometimes it would go for hours without failing. Sometimes even a slight voltage bump would cause it to crash instantly. If I go back to the settings that worked fine earlier, it would crash again.

Last night I decided that it was high time that I moved away from Prime95 and found something new that worked well with Haswell. I have tried AIDA64, but people have been complaining that even after passing it for 12Hrs, it would crash in games. I don't know if that is a good stress test either. Plus, you have to pay for it.

Some guy one Overclock forums recommended that he is using Intel's own stress testing utility called Intel Extreme Tuning Utility (a.k.a Intel XTU). People say that it fully supports the new AVX2 instructions, doesn't make the Vcore go out of control when the CPU is bombarded with AVX instructions and it also shows all the voltages and temps on the same screen (so that you don't have to run a separate monitoring utility (or utilities) as is the case with Prime95. AIDA64 also shows the voltages and temps while stress testing. However, XTU also allows you to overclock the CPU from itself. But I don't really like overclocking using software, especially when AISuite III is installed there and it can clash with other software.

So, I started a stress test in XTU last night and I ran it for 13hrs without a crash or error. However, I'm not confident that not failing XTU alone is a good indication of stability. So I am running a few hours' long H.264 transcode queue in HandBrake. If I can get through it without an error, then I guess I would be stable in everything that I do. I only stress tested the CPU portion. I skipped memory for now. If H.264 fails, I would test Memory as well. Otherwise I don't see why I need to do that because the clocks, timings and voltage is pretty standard for today.

The current overclock settings stands as follows.

• Core multiplier: 44x (sync all cores enabled)
• Vcore: 1.285V manually set in BIOS (should change to Adaptive later)
• Un-core multiplier: x39 (AUTO)
• Un-core voltage: 1.15V (AUTO)
• RAM speed: 1600MHz (XMP)
• RAM timings: 9/9/9/24/2T (XMP)
• RAM voltage: 1.50V (XMP)
• Input Voltage: 1.75V (AUTO)
• Extreme Tweaking: Disabled
• EPU Power Saving Mode: Disabled
• Everything else: AUTO

※This is on hte Asus Maximus VI Hero board with BIOS ver.711.

Here's proof.

Note: 
Time remaining says 10hrs 58min because I set it to 24hr stress test. Deduct that from 24hrs, and you get 13hrs 2min. 

[Article] I give up. Have to settle at 4.3GHz with my 4770K.

In a previous post I mentioned that I was able to clock my new Haswell Core i7 4770K chip to 4.3GHz and I was planning to clock it to 4.4GHz. After trying for a week or so, I couldn't get it Prime stable at 4.4GHz below 1.3V Vcore. My CPU is a dog. It doesn't overclock well at all. In fact, it is a well below average chip. That's sad. I lost the silicon lottery.

But it is not the end. There is a few more tweaks I can do. For example, right now the RAM is at 1600MHz with XMP timings (9/9/9/24/2T/1.5V etc.) and cache ratio is at 39x with 1.075V. I think I can tweak the timings a bit and increase the cache ratio to 41x or so. 43x is ideal but will be a long shot  I guess. I would definitely have to increase the Un-core voltage to somewhere around 1.2V to get near 4.1GHz. But that doesn't add a lot of power so I should be alright. Maybe I could tweak the voltages a bit. Perhaps lower the CPU input voltage to cut the temps by a degree or two. I could even play with the BCLK as well. Perhaps put it to 105MHz with the multiplier dropped to 41x? That might improve the performance slightly. Something is better than nothing, right? 

Anyways, I am really disappointed with my CPU. There are people who have chips than can clock to 4.9GHz with just 1.25V. I have a great motherboard - a motherboard made for overclocking. But me not being able to get higher than 4.3GHz overclock is not the board's fault. It's the CPU alone. I probably can get by with 4.4GHz at something like 1.29V and be stable in real-world work loads. (Heck, I played Crysis 3 campaign from beginning to end with just 1.28V and didn't encounter a single crash). But in my head I know that it is not 100% stable and it might crash when I least expect it to. That is not something I can live with. Hey, it's just 100MHz which is even less than 2.5%.

One thing I should tell is that it was almost always the in-place Large FFT test that crashed when I tested for stability at 4.4GHz. Playing with the voltages other than the Vcore didn't made it more stable. It just needed more Vcore. When I kept on increasing the Vcore, the duration until Prime95 crashed lengthened, which means it was Vcore starved. 1.3V is a lot of volts for Haswell, no thanks to Intel cheaping out under the hood. Bastards!

Funny thing was that I could run 10 passes of Intel Burn Test at very high memory usage at 4.5GHz with just 1.280V Vcore. Intel Burn Test might be an indication of stability, but it is way inferior to Prime95, even though the temps while running Intel Burn Test is much higher than when running Prime95.

Here's are my current settings for 4.3GHz.

• Vcore: 1.235V manually set in BIOS (1.248V max recorded in AISuite III)
• Un-core multiplier: x39 (AUTO)
• Un-core voltage: 1.075V manually set
• RAM speed: 1600MHz
• RAM timings: 9/9/9/24/2T
• RAM voltage: 1.50V
• Extreme Tweaking: Disabled
• EPU Power Saving Mode: Disabled
• Everything else: AUTO

Now, I haven't run adequate stress tests. I will do that after tweaking the un-core and the memory timings a bit.

[Article] To de-lid my Haswell 4770K CPU or not?

Haswell runs hot compared to SandyBridge. IvyBridge also ran hot, but Haswell runs even hotter than that because of the existence of FIVR (Fully Integrated Voltage Regulator). SandyBridge CPUs had the heatspreader soldered to the die thus having good heat transfer from die to the heatspreader. IvyBridge and Haswell both don't have it soldered. Instead, they have a thermal paste in between the die and the heatspreader. Intel is cheaping out.

The thermal paste itself isn't the issue. The biggest issue is that the "large" gap between the die and the heatspreader that requires a thicker than ideal layer of thermal paste. This gab is caused by the thick layer of glue that holds the heatspreader in place.

The only option is to either lower the temps by lowering voltages thus lowering the overclock OR to de-lid the CPU. That means, you pop open the heatspreader, scrub off the glue, clean off the thermal paste, put a better thermal paste and putting back the heatspreader. If properly done, people have been seeing up to 30°C drop in temps after doing this. THAT is massive.

Lower temps help overclock higher, or run your fans at a lower RPM. But it also lowers the power consumption of the CPU. What? How's that possible? It has something to do with these Tri-gate transistors. More voltage obviously increases power consumption. Higher temps also increase power consumption. Well, if you use power saving features - and your should - this will only happen at full load.

I compiled a list of FAQ below and will keep adding more stuff as time goes by.

[Article] Overclocking my 4770K – breaking 4.3GHz

You must be wondering why I want to overclock a super fast CPU like the 4770K. Fun, man, fun! No other reason. All the chips that I have had so far, maybe not my Core i5 750 which clocked +50% at stock volts, have been poor to average clockers. My previous 2600K CPU needed about 1.34-1.36V for 4.5GHz.

Unfortunately, my 4770K chip doesn’t break that spree. When it comes to Haswell, there is a simply criteria for checking whether the CPU is a good one or a bad one. If you can boot to Windows with 4.6GHz with just 1.2V, then you have a very good CPU. If you can do that with 1.25V, still your chip is above average. So the first thing I did was to set 4.6GHz and 1.2V in BIOS and check if it can boot into Windows. Nope, it couldn’t. I was greeted with a BSOD just before Lock Screen. I checked 1.25V. Still no go. Dreams shattered!

So I set 4.2GHz and 1.2V and slowly went up. With a lot of testing, I found out that 4.3GHz at 1.24V was stable enough for my liking. This is not the end of everything as I had to lower the cache ratio to 35x and memory clock to 1333MHz SPD. That sucks, but none of them can match the benefits of a 100MHz overclock on the core. Besides, you can increase them later and find the sweet spot. Right now I am only concerned about core clock. 4.5GHz is my ultimate goal, but from the way things are going, I might not be able to get there without a de-lid.

To summarize, here are my 4.3GHz stable settings.

• Vcore: 1.240V manual
• Uncore multiplier: x35 (min = max = 35)
• Uncore voltage: AUTO (1.05V)
• RAM speed: 1333MHz
• RAM timings: 9/9/9/24/2T
• RAM voltage: 1.50V
• Extreme Tweaking: Disabled
• EPU Power Saving Mode: Disabled
• Everything else: AUTO

With the above settings, I managed to pass

• Prime 95 v27.9 blend test for 6Hrs
• Handbrake H.264 encoding for 2.5Hrs
• Crysis 3 Multiplayer gaming for 2Hrs
• AIDA64 stress test for 1Hr.

That’s stable enough for me. I don’t want to run Intel Burn Test or anything because if it passes Prime95, that’s all I need. The others are just for insurance.

Right now I’m testing 4.4GHz. I had to bump the Vcore to 1.28V. It passed AIDA64 for 2.5hrs. I’ve been running H.264 encoding for almost 2hrs. I haven’t been able to run Prime95. For some reason, when I ran it, the core clocks were fluctuating between 4.2GHz and 4.4GHz. I couldn’t understand why that was. Maybe some software glitch. I will check it after I finish with the H.264 encode. Maybe a simple reboot with fix it.

Hope it will be stable after I fix Prime95 issue. Wish me luck.

[Article] Upgraded the rig. Haswell it is.

So, I moved to Haswell after all the bad things I said about it. This was the primary reason, but I wanted to try out something new as well. Right now I'm trying to find the best overclocked settings for the CPU and it is hard work.

I only bought the CPU and the motherboard, but I made some other changes as well. Replacing the X-Fi Titanium with onboard SupremeFX audio and removing the fan controller are the two main things.

To the newest specs then.

CPU: Intel Core i7 4770K @4.4GHz / 1.280V in BIOS (still testing)
CPU Cooler: Thermalright SilverArrow SB-E Extreme
Motherboard: Asus ROG Maximus VI Hero
RAM: 16GB Corsair Vengeance CMZ8GX3M2A1600C9 9/9/9/24/1T/1.35V
Video Card: Palit GTX670 JetStream @1056MHz (base clock)/6800MHz RAM
SSD: Samsung 840 series 250GB
HDDs: Toshiba DT01ACA300 3TB, Western Digital Caviar Green WD30EZRX-00DC0B0 3TB
Display: Iiyama Prolite E2773HDS
Case:  Silverstone Raven RV-03
PSU: Corsair TX850
Mouse: Logicool G500
Mouse Pad: Steelseries QCK / Razer Sphex
Keyboard: Logitech G15 V2
Headset: Roccat Kave 5.1
Speakers: Logicool Z623
Webcam: Logicool C910

Still there is a lot of things to check with the new setup. I've been trying to get my overclock stable. Currently trying 4.4GHz. The temps go up ridiculously when you stress test with Haswell CPUs, and I have to use a different method to stress test. So it is taking a bit longer.

Cheers.

[Article] Haswell overclocking woes - the deal with Vcore, IVR and AVX instructions

In the previous article I discussed what the main causes of high temps on Haswell CPUs. Now let's look at another issue.

When you overclock your CPU, there are two main method how you would set the Vcore. Static mode and Offset mode.

Static mode

Static mode means that even when the CPU is idle or doing light work, the Vcore would remain as if you are doing heavy work. The Vcore would not change dynamically according to the work load. Now, there is no real need for that. Lower Vcore is always better for longevity of the chip.

Offset mode

Thus there is Offset mode. Intel CPUs automatically can set a Vcore based on the multiplier by referring to voltage tables embedded in the chip. These voltages are called VIDs and higher the multiplier, higher the VID. But Intel can only guarantee the VID is stable only within the specifications. After that, if can be either stable or not. At some point you'll find out that the stock VID is not stable enough. Thus you need can specify an offset to the dynamic VID. It doesn't have to be positive. If you know your CPU can work at a lower Vcore than the VID at max multiplier, you can specify a negative offset. But there is one issue. The offset applies to entire multiplier range. That means, even at idle, the offset value would be applicable. Especially if you are using a negative offset, it can crash the system. Or it won't be optimal.

So with Haswell, Intel introduced a 3rd method. Adaptive mode.

Adaptive mode (Haswell only)

Adaptive mode works like Offset mode, but only applies to turbo multipliers. That means, the dynamic Vcore would be identical to the VID below turbo frequencies. Thus, this method does not cause instability at idle states due to negative offsets an does not over-volt within the specified multiplier. Pretty good deal.

Except, it is not. Actually, it's not Adaptive mode's fault.

I'm sure that you already know that there is an IVR onboard the CPU package. The IVR is not the entire CPU voltage regulator. The old VRMs are still there on the board and are needed. The IVR is there to control the Vcore more aggressively. Basically, the motherboard inputs the CPU a quite large voltage called the Input Voltage. The IVR give the cores a portion of the this Input Voltage. It can rapidly fluctuate voltage depending on the load and multiplier.

OK, now that looks like a nice thing. But AVX instructions make the Vcore go crazy. I don't know why, but for some reason, whenever the AVX instructions are used, the IVR increases the Vcore by about 0.1V. This happens in Offset mode as well. And there is nothing you can do about it other than use static Vcore. With static Vcore, you don't have this problem. Even when AVX instructions are being used, the Vcore doesn't change.

Stress testing with AVX loads is the problem

This doesn't look as bad as it sounds in reality because you might not use AVX instructions at all. Plus, the increase in Vcore doesn't necessarily increase the temps in normal work loads as if you are running a stress test. If you run Prime95 or LinX or AIDA64 which stresses the CPU with AVX loads, it would definitely make the temps skyrocket. For this reason, they say not to stress test using Offset or Adaptive methods. Old Prime95 (pre 27.7) is OK because it doesn't support AVX. Now, one option would be to run at Offset mode or Adaptive mode and then switch to static mode once you start using AVX supported applications, if you ever. The upcoming H.265 encoding is supposed to make use of AVX instructions, and that would be a problem for Adaptive mode or Offset mode. I'm hoping to move to H.265 once it becomes stable. You get the same quality for half the file size. That's something you don't wanna miss out on.

There shouldn't be a big difference in power consumption between static Vcore mode and either of the dynamic Vcore modes as long as you are using all the power saving features. In fact, some say (I haven't been able to confirm it) that on the Asus Z87 boards, at full idle, the Vcore drops to idle Vcore levels as if it is using Offset mode. This seems to be how Gigabyte boards operate. I'm gonna check this out when I get back home. I think the board has the C6/C7 power states disabled by default. My PSU should be able to handle those low currents.

[Article] Haswell overclocking woes - Super high temps and why that is

In the last post I mentioned that I bought a new Haswell processor - the top of the line Core i7 4770K - and a board with the Z87 chipset - Asus ROG Maximus VI Hero. We all know that retail version of Haswell CPUs don't overclock well, and luck plays a big role when it comes to overclocking. It was kinda like that in SandyBridge and IvyBridge as well, but now the weight has moved even more towards the luck factor. I wanted to believe that I was on that top 5% of the crowd that were lucky enough to get a chip that could boot into Windows at 4.6GHz with just 1.2V. Well, it turns out that I am not that lucky. But my chip isn't that bad either.

I managed to get mine up and running at 4.4GHz on all cores, 4.1GHz on "uncore", with a Vcore of 1.285V set in BIOS. Oops, sorry, UEFI. That Vcore doesn't sound too bad for SandyBridge, but for Haswell, it is a big Vcore. Not electrically, but thermally. Usually, this is the custom water cooling territory. However, the CPU managed to pass 6.5hrs of AIDA64 stress test overnight. The max temps it hit was 89C, but that's only briefly. Most of the times it was below 80C. My Silver Arrow SB-E Extreme was doing a pretty sweet job it seems. And that's not even at 100% fan speed. Just 1600RPM on average. I can go all the way up to an ear deafening 2500RPM if it deems necessary.

Reasons for high temps

First let's deal with the obvious stuff. Why does Haswell CPUs run hotter than their predecessor? There are three reasons. 

• It uses Tri-gate transistors that don't like to scale with voltage. We saw this with IvyBridge as well
• There is a voltage regulator module onboard the CPU, called the IVR (Integrated Voltage Regulator). That's additional circuitry and circuitry that handles voltages.
• The die is not soldered to the heat-spreader. There is merely a thermal paste in between the die and the heat-spreader. And a very bad one at that.

Hence, as you increase the voltage when you are overclocking, the core temps go up rapidly. Don't be mistaken though: it's not heat. The heat output should be less than SandyBridge and about the same as IvyBridge. It's the temps that go out of control because the heat is not properly transferred from the die to the heat-spreader.

Solution

What's the answer? There are two. Neither of them are happy solutions though. 

First one is, lowering the Vcore. You will have to give up your current overclock. That's definitely not a happy solution. 

Second one is, delidding. Delidding means, you remove the heat-spreader, clean out the crappy thermal paste, put high quality thermal paste and putting back the heat-spreader. That's a bit adventurous and definitely voids warranty (and perhaps you won't be able to sell it back), so again, it is not a happy solution either. But if properly done, you would see a drop of about 15C by delidding. That's a huge amount. Even if you don't exploit that added thermal overhead, you will have a cool running CPU that even uses less power. (Heat causes more resistance and thus increases power consumption. On IvyBridge that was like a 30W difference at 4.6GHz. It should be similar on Haswell.)

I haven't decided if I want to delid or not. That's a big decision to make. But I will talk about how to do it on another post.

[Article] Haswell, here I come: 4770K + Maximus VI Hero.

Two years and one month ago, I got my SandyBridge CPU (2600K) and motherboard. Yesterday, my motherboard broke. Well, after a long sleep, it came back to life. Still not completely stable, but it is working at least. But I decided that I’m gonna get a new motherboard anyway. It’s a matter of time this board will die imho. I just hope I will be able to sell the motherboard.

I had been sitting on the fence in dilemma, uncertain whether to stick with my SandyBridge 2600K CPU or to upgrade. I first felt Haswell was not worth it. Then the feature-set of the new boards won me over. I even considered the upcoming IvyBridge-E CPUs. But they will be using the two generations old x79 chipset. The prices would be pretty steep (3930K costs JPY57,000 even now), so I decided that route is not for me. That's JPY80,000 or more.

I had been reading the motherboard reviews to decide on a board. Three boards came out on top. The Asus Z87 Pro, Asus ROG Maximus VI Hero and MSI Z87-GD65 Gaming. I felt the software/UEFI support of Gigabyte boards was cheesy and ASRock boards not so good in terms of power delivery circuitry, software and even price - which was their strong point until now.

[Article] Upgrade itch is so strong! Feeling like upgrading to 4770K Haswell from 2600K SandyBridge after all.

Many shops are giving JPY5,000 worth discounts if you buy a Haswell CPU and a Z87 motherboard these days. Sofmap.com has a eye-catching offer: Core i7 4770K and a Asus Z87 Pro motherboard for about JPY51,000. I sometimes feel like giving a shot at Haswell. I know it won’t give me much benefits over my current CPU and motherboard. But this upgrade itch has been crawling beneath my skin in the past few months. Worse case would be that I will get just about same performance as my current CPU and motherboard combo. That’s if I can only overclock to like 4.2GHz. If I can somehow manage to overclock to 4.5GHz, I will be able to sneak in 10% or so more performance. The temps definitely will play a vital role, especially with the temperature in the summer.

It’s not the CPU that is great actually. Z87 is a great chipset. It performs really well, plus, it uses so little power. It also comes with a lot of stuff my current motherboard doesn’t have, natively. I will get a much better motherboard. I can use the integrated fan controller and throw away (don’t take things literally unless I say so) my current Scythe fan controller. I have been wanting a fan controller that controls fan according to temps. With Asus’s unprecedented fan control software, I can achieve it. I hate the cable mess the current fan controller has created behind the motherboard tray.

Is it worth upgrading to Haswell from IvyBridge or Sandybridge?

The official Haswell reviews came out at the beginning of this month (June, 2013) and they proved that the information that was leaked few weeks ago is completely accurate. Haswell is only having a single digit (percent) performance advantage over IvyBridge and they get very hot when overclocked. It seems you need a lot of luck and cooling power to even hit 4.4GHz. However, what you have to do to overclock the Haswell CPU should be quite simple.

Those new z87 motherboards are so delicious though. You get 6 SATA-III ports, 6 USB3.0 ports, PCI-E 3.0, integrated Wireless 802.11ac (certain models), much better integrated audio (certain models) and nice looks overall. Some people would even upgrade to Haswell for this alone. 

Check this HARDOCP article to see how Sandybridge, IvyBridge and Haswell all stand against each other when overclocked to 4.5GHz. This article also mentions all the pitfalls of overclocking Haswell. It's a good read. You should read it if you are planning to get a Haswell CPU and want to overclock it.

BTW, if you are planning to get a non-K CPU hoping that it would allow you to overclock at least a few hundred MHz, you would be disappointed. Only K-editions are overclockable. Others will be stuck to their stock Turbo Boost algorithm. You can try increasing the BCLK, but it won't get your much farther. 

On top of that, you don't get VT-d and TSX instructions support on the K-editions. They are only available on non-K editions. Intel is messing with the Overclocking community it seems. It clearly shows that they don't want use to overclock, because it handicaps their sales because people might not upgrade if their current CPU overclocks pretty well and is faster than the new CPUs. (This seems to be the case with a lot of SB and IB users against Haswell)

I evaluated a bit where the economics stand. I would obviously sell my existing CPU and motherboard if I upgrade to Haswell. According to the current second hand prices, I would only get like ¥22,000 for both of them. I might even add my Creative X-Fi Titanium audio card to that so I might get like ¥24,000 (because integrated audio on these boards is so good). But the Core i7 4770K alone costs little more than ¥35,000 and then add another ¥20,000 for a good motherboard. (It's the massive price you have to pay for getting stuff at launch here in Japan. Prices will drop few thousands in the coming months.) That's about ¥30,000 extra for almost no performance increase (well 15% best case increase isn't a lot). I must be insane to go ahead with the upgrade. 

Thus, my upgrade plans are officially done. I'm keeping my trusty Sandybridge. But being the geek I am, I won't stop there and give up. There are newer CPUs coming in few months, from both Intel and AMD. Hopefully I will be able to upgrade to one of them.