DDR4 Info:D?

[deathjester]

1. Ivy Bridge is well known for being more resistant to voltage degradation than Sandy Bridge.

 

2. Sin has been very knowledgeable about overclocking and power delivery, from everything I've seen from him.

 

3. Asus ROG boards, apart from the extreme, had 4 phase PWM drivers (with doublers so they could advertise an 8 phase) on Z77 so I'd assume they're still going for as few phases as possible, and doublers halve the PWM frequency, so they don't increase voltage stability.

 

4. MB manufacturers are going to be very cautious about the maximum temperatures and voltages they give people because they don't want people to go and accidentally kill their parts. The recommended max that Sin gives is the one that you should never exceed (IIRC, he gave 1.52V as a max for sandy) whereas the recommended maximum by the manufacturer is the voltage which you can exceed but are recommended not to. This would be something like the maximum recommendation of 1.35V on Sandy. 

 

5. Delidding Haswell can lower temperatures by a significant degree, since it's not soldered like Sandy Bridge, which will buy you more overclocking headroom. CPUs would seem to be getting faster if Intel was not for this.

 

 

Do you think anyone who actually owns a Haswell believes the hilarious "tech advice" you linked from "Sin at overclock"? 1.45v on AIR on a Haswell? HAHHAHAHHAHAAAAAAAAAAAAAAAAAAAAAAAAAAHAHA.

 

You really seem to have an agenda here. Do you work for Intel or own shares by chance? I really grow tired of these shenanigans while trying to help people on a forum. If you go over 1.4v temps will not be ok, not on triple rad with a delid. The only way they were ok is on engineering chips that we don't have and that ran cooler. They were probably soldered to the lid and the highest bin possible out of thousands of parts.

 

You have a 2600k and are pushing Haswell as a product which makes no sense at all. All your posts reek of "selling" a product and you use articles and stats based on engineering chips that were shown to be completely inaccurate compared to retail chips. You then  use " internet personalities" that probably couldn't graduate from ITT tech and were most likely using an ENGINEERING SAMPLE to test with in the first place.

 

Please anyone on this forum running with a RETAIL chip run your 4770k at 1.45v on air. Link a screenshot with cinebench and temps with CPU-Z open or better yet a real render with Asus Real bench and just h.264 clicked. I can't wait to see those temps. They should be GOOD TO GO, per "sin" from overclock forum. 

 

I am done arguing this. If people actually believe these lies, and outright DANGEROUS advice you gave them? They deserve to break their chip. I have seen and overclocked like 50 Haswell's and mine was among the best and can't come close to doing 4.7 on air SAFELY. 4.5 at 1.22v only works because my chip for some reason can get by with 1.777 VCCIN, when most chips need 1.9 to hit the same clock. It saves me just enough on the temps or I would be running at 4.3, since 4.4 and 4.5 need almost the same voltage. My chip will do 4.7 (anything past that would be with cores not at the same speed or HT off which is pointless) SAFELY on water for a 24/7 system and I haven't come across another one that would.  I consider myself LUCKY and at 4.5 I am still losing to an OC Sandy on the forum cincebench thread in single thread performance, who is using a stock cooler and prob has lower clocked ram. At 4.6 I am losing to him by one point and would need water cooling. At 4.7 I would match his single core performance on dual/triple rad, while he could run the OC with an EVO 212 with lower temps...

[Art Vandelay]

Do you think anyone who actually owns a Haswell believes the hilarious "tech advice" you linked from "Sin at overclock"? 1.45v on AIR on a Haswell? HAHHAHAHHAHAAAAAAAAAAAAAAAAAAAAAAAAAAHAHA.

 

You really seem to have an agenda here. Do you work for Intel or own shares by chance? I really grow tired of these shenanigans while trying to help people on a forum. If you go over 1.4v temps will not be ok, not on triple rad with a delid. The only way they were ok is on engineering chips that we don't have and that ran cooler. They were probably soldered to the lid and the highest bin possible out of thousands of parts.

 

You have a 2600k and are pushing Haswell as a product which makes no sense at all. All your posts reek of "selling" a product and you use articles and stats based on engineering chips that were shown to be completely inaccurate compared to retail chips. You then  use " internet personalities" that probably couldn't graduate from ITT tech and were most likely using an ENGINEERING SAMPLE to test with in the first place.

 

Please anyone on this forum running with a RETAIL chip run your 4770k at 1.45v on air. Link a screenshot with cinebench and temps with CPU-Z open or better yet a real render with Asus Real bench and just h.264 clicked. I can't wait to see those temps. They should be GOOD TO GO, per "sin" from overclock forum. 

 

I am done arguing this. If people actually believe these lies, and outright DANGEROUS advice you gave them? They deserve to break their chip. I have seen and overclocked like 50 Haswell's and mine was among the best and can't come close to doing 4.7 on air SAFELY. 4.5 at 1.22v only works because my chip for some reason can get by with 1.777 VCCIN, when most chips need 1.9 to hit the same clock. It saves me just enough on the temps or I would be running at 4.3, since 4.4 and 4.5 need almost the same voltage. My chip will do 4.7 (anything past that would be with cores not at the same speed or HT off which is pointless) SAFELY on water for a 24/7 system and I haven't come across another one that would.  I consider myself LUCKY and at 4.5 I am still losing to an OC Sandy on the forum cincebench thread in single thread performance, who is using a stock cooler and prob has lower clocked ram. At 4.6 I am losing to him by one point and would need water cooling. At 4.7 I would match his single core performance on dual/triple rad, while he could run the OC with an EVO 212 with lower temps...

I trust someone who actually has experience overclocking more than "rog editor" at Asus. 1.52V on Sandy Bridge sounds insane, but that's the NEVER EXCEED voltage, not the maximum realistic safe voltage.

 

That's pretty tinfoil hat to accuse me of being from Intel.

 

I used a thread from overclock.net where people who were not sponsored by anyone posted their overclock results, and I used an article written by an experienced overclocker. You gave me an article written by a motherboard manufacturer that obviously doesn't want people having failed parts.

 

Once again, you're completely misinterpreting the value of 1.45V and the meaning of the word "air". By "air" he meant cooling that doesn't go below ambient (at least I'm assuming that since he didn't have a section for water). A water cooling loop would fit that description. I'll try and dig up some screenshots from the Haswell overclocking thread on OCN, since they have a bunch of people using high voltages on ambient cooling.

 

When you overclock, you take the risk of breaking your chip. Frankly, anyone who is exceeding 1.35V on a Sandy Bridge chip is slowly killing their CPU, but plenty of people run above that. I can't run above 4.5GHz within that "safe" voltage range.

[Art Vandelay]

This guy managed to run the chip at 1.35V on ambient cooling.

 

This wasn't the most amazing overclock, but it was done on a 240mm rad and is at 1.312V. Not even close to needing a triple rad.

\

[deathjester]

1) I could care less what a delid did since most people aren't delidding.

2) 1.312 isn't close to 1.45v, that is close to 1.3v which is what Intel recommends as a MAX.

3) 80C on cinebench is in no way safe or acceptable temps to run as a 24/7 machine that will be rendering or streaming/playing games. He is about 8 degrees C over acceptable temps. Blah blah it is only 8 degrees. Water boils at a certain point. Chips degrade at a certain point. Running past that point isn't good.

4) Hyperthreading adds about 10 degrees so that chip would be at 90C if it was a 4770k and good luck getting a 4770k to 4.9 ghz with HT on at anywhere close to 1.3v.  

5) Running cinebench and Asus stability test once doesn't mean jack as far as stability. 

6) this is prob a NH-D-14 which is as good or better then cheap water kits. The cooler is fantastic and that is why it costs almost as much as an H100i

 

See the "ES" in CPU-Z in this and all initial BS reviews of Haswell? That means Engineering Sample. When you see that in CPU-Z? Those results are a joke. Even the engineering samples could pass aida for hours on end and fail true stability in real applications.

http://www.hardocp.com/article/2013/06/01/intel_haswell_i74770k_ipc_overclocking_review/7#.UsTHAfRDu30

 

Thank you for posting a pic of someone running a non HT chip past it's thermal limit at near 1.3v though when it is probably delidded. It really lends a lot of credibility to 1.45v on air that "sin" suggested...

[SoaresGaming]

I think i need to unfollow this topic..  :blink:

[ViruZ_HUN]

I think i need to unfollow this topic..  :blink:

me 2

[LukaP]

you guys are insane. first off, of course Moore's law still stands, getting IPC improvement combined with more cores equals better performance

 

second die shrinks....

 

@deathjester haswell didnt shrink die. tocks do that. and yeah ivy does run hotter than sandy. know why? because solder was replaced by a thick layer of TIM, just like in haswell.

 

and yeah you get variable results, but AVERAGELY the speed is increasing. just check independent threads @Art Vandelay posted. 

 

so stop your mindless hate on Hasswell just because your chip didnt go to 4.8 @1.2V... 

 

if you were to delid, you would see a temp drop of 10-20C (i got 17, seen more than 20 even) therefore getting your haswell into safe temps for higher voltages, basically making it perform very similarly to a SB, except IPC improvement and a die shrink which does admiteddly produce a little more heat.

 

so, to conclude Haswell is still stronger than SB, and it will continue to improve, with BW bringing 14nm and an of die VRM it means similar heat, better IPC, and mabybe solder under the IHS. ergo, faster chip. ergo, Moore still stands

[ViruZ_HUN]

So if anyone else knows something else about DDR4, feel free to write it down.

 

I dont want to hear more about CPUmasterrace

[deathjester]

So if anyone else knows something else about DDR4, feel free to write it down.

 

I dont want to hear more about CPUmasterrace

No one knows anything except it uses less voltage and starts at 2133mhz. We do not even know if it is faster (lower latency). At this moment it appears just to be a "green pc" part, since they would have been bragging about it's latency if it was anything to write home about.

 

you guys are insane. first off, of course Moore's law still stands, getting IPC improvement combined with more cores equals better performance

 

second die shrinks....

 

@deathjester haswell didnt shrink die. tocks do that. and yeah ivy does run hotter than sandy. know why? because solder was replaced by a thick layer of TIM, just like in haswell.

 

and yeah you get variable results, but AVERAGELY the speed is increasing. just check independent threads @Art Vandelay posted. 

 

so stop your mindless hate on Hasswell just because your chip didnt go to 4.8 @1.2V... 

 

if you were to delid, you would see a temp drop of 10-20C (i got 17, seen more than 20 even) therefore getting your haswell into safe temps for higher voltages, basically making it perform very similarly to a SB, except IPC improvement and a die shrink which does admiteddly produce a little more heat.

 

so, to conclude Haswell is still stronger than SB, and it will continue to improve, with BW bringing 14nm and an of die VRM it means similar heat, better IPC, and mabybe solder under the IHS. ergo, faster chip. ergo, Moore still stands

 

Show me a 4770k that wasn't an engineering sample that does 4.8 @ 1.2v. At 1.2v your heat would be ridiculously low. The person should be able to run freakin prime blend and show a stability test on it. The only person I see on this entire forum who got 4.8-4.9 thought well over 1.4v was good for voltage. Guess what they run their chip at 24/7. It sure as hell ain't 4.8 ghz...

 

We have multiple posts on this board with people who need 1.3v on a I5 Haswell to be stable at 4.2ghz. Another one was just posted today. Why do they think their is something wrong with their chip? Because of BS engineering samples that people think are retail chips. All these lies do is tick people off. Thanks for being a part of that!

 

Most people don't care about delidding on Haswell for a few reasons. 1) It voids your warranty. 2) You can delid a 1.3v 4.2ghz haswell and all you are gonna do is run unsafe voltage at 1.4v and kill your chip for maybe 4.3? Great advice. Delidding is freakin pointless when the chips are just BAD. You can't send the voltage through Haswell that you could on even Ivy. So what is the point? The gains are minimal and the risks are big. You trade dangerous voltage for bad temps for 100 more mhz? Are you advising them to send their chip back to intel under warranty it delidding breaks it and not tell Intel? That is fraud. Might as well tell them to rob a liquor store and go all out. Their screw ups make the chips cost more for everyone. If you want to risk 300 plus bucks which is the average cost of a 4770k? It should be your money at risk.

 

When OC chips = same that means thermals have taken control and we aren't going anywhere. What does your Haswell run at buddy? Please post a screenshot with temps, cinebench and cpu-z open...

 

A turbo little car is gonna beat a v8 clunker car in a race. The average PC Sandy beats out multiple people who have got stuck at 4.2 in the past few days. Those people are all lying though right? 

 

Please post a pic of your 4.8ghz 1.2 v 4770k (LOL) with a something like cinebench running and HW monitor open. I am betting you have adaptive voltage on and have no idea how much voltage is even running through your chip... 

 

Thanks for playing though. The other guy already left the thread after he thought 1.45v on AIR was a good voltage for Haswell, which might have been the most ridiculous post I have ever seen.