Just in case 
Overclocking "Raptor Lake refresh" - 14900k,14700k, 14600k etc results, bins and discussion
33 reading
Nizzen
16,221 - 16,240 of 23,546 Posts
Joined
·
2,973 Posts
This is his VMIN for 5.9 at 33c liquid and on direct die. I said that is 70mv worse compared to my chip at the same temps(extrapolated)
He is arguing that his liquid temps are worse and if it were lower like mine( my liquid temp was at 20c ), it would be closer to mine. I told him that it would but then you will have over 10 degrees lower temps than i do because you are on direct die which would not be a fair comparison.
1.275V load vmin at 88C (33c liquid) { VOUT }
Here is mine
1.205V load vmin at 75C (20c liquid) { VOUT }
1.23v load VMIN at 86C (26c liquid) {VOUT}
Joined
·
362 Posts
The video with the voltage claim was "Rambling about the new Intel 13th/14th gen Intel recommended default settings."
Commenter said
Commenter said
Where would worst sample VID (with TVB voltage optimization) vs their ratio clipping temperatures end up if plotted?
Joined
·
721 Posts
In some of the screenshots posted by other users I've seen 1.484V for 6.2 GHz , and I think the clipping temperature is at 70 °C. TVB optimizations at least on my 12th gen CPU decrease voltages by 1.5 mV / °C (but this might change depending on the SKU, and the equation is not exactly linear).
1.5V at 70 °C with TVB optimizations could be about 1.45V, so right at the edge of what would be continuously allowed.
You are more than welcome to send me your chip to test out but need to delid it 1st
whats this about violating Intel spec , is this not OCN where things are pushed a bit beyond their limit or now its all within " Intel Spec " 😔 As mentioned i will 100% run the test again and even rerun the 5.9 but when temps are somewhat more realistic with how i normally test which is in the mid 20's WT. It was mention previously that all the higher SP fall inline within one another by 10mv or so at 6ghz mine included and i specifically said its a lot closer than the V/F would have you believe
Joined
·
292 Posts
Not linear: progressively higher voltages would require even lower temperatures, don't know how it would behave when approaching 0K(irrelevant from a engineering point though). The numbers from that comment look solid: last year Derb8uer in an interview with an Intel engineer in his Youtube channel asked him about this and he told that they usually set for voltages of 1.2-1.3V for heavy loads and that throttling in those conditions was fine(100C). So the 1.3V@100C voltage reliability is good.
I would picture it drawing from the comment as:
1.3V@100C
1.45V@70C
1.6V@30C
1.75V@-20C
1.9V@-80C
2.05V@-150C
Temperature slopes would get bigger for the same voltage slope as voltage increases: for practical real world use the slope increase could be approximated as linear.
its mounted as good as it can be and core to core delta is around 9 deg ( not the best but not the worst ) , i suppose it would be good to see another iceman user post temp who is not using a chiller basically similar setup to mine to see where there delta is at . As i have mentioned previously my water at 33deg is doing me no favours but is what it is without a chiller and going into summer , room temp is up at 28 deg .... everything is just hot 😂
Water Temp is Key ... 23 deg water temp with 380W with 55deg delta 1.288Vcore min for R23 multi runs
Water temp at 30 deg with 400W and 60 deg delta but needed more vcore at 1.314vmin to pass
Joined
·
721 Posts
Interesting. From the data points, like this, then?
Although for a more practical temperature range it will be like this:
I imagine that people who deliberately (or not, due to motherboard default settings) disable TVB might be at risk of long-term degradation in some usage scenarios.
This seems important information to keep in mind when overclocking, it's a pity that it isn't in the public datasheets.
It just didn't seem like from the #'s you posted (post I quoted before) that your delta is low enough. Should be 10-20C lower, at least based on what others say in regards to the temp drop from going DD. Either that or I need to buy an Optimus block.its mounted as good as it can be and core to core delta is around 9 deg ( not the best but not the worst ) , i suppose it would be good to see another iceman user post temp who is not using a chiller basically similar setup to mine to see where there delta is at . As i have mentioned previously my water at 33deg is doing me no favours but is what it is without a chiller and going into summer , room temp is up at 28 deg .... everything is just hot 😂
Water Temp is Key ... 23 deg water temp with 380W with 55deg delta 1.288Vcore min for R23 multi runs
![Image]()
Water temp at 30 deg with 400W and 60 deg delta but needed more vcore at 1.314vmin to pass
![Image]()
Joined
·
721 Posts
This is close to some of the worst 12900KS samples I've seen screenshots posted of in the past.Worst case is quite a bit higher.
![Image]()
I'm not aware here if Enhanced TVB with 13/14th gen CPUs scales the TVB cutoff temperature dynamically or if TVB optimizations reduce voltages on the best cores more aggressively with these CPUs. If not, then Intel might be indeed having default VIDs for some processors that can cause long-term degradation (if the previously posted information on temperature vs voltage is generally valid, that is).
Joined
·
362 Posts
Is 70C where the single core 6.2 is clipped? If so, is that where it is supposed to be clipped? I am too ignorant of how the Raptor i9 are supposed to work, 12th i7 clock ratios are pretty simple.
On Zen 3, besides the current, power, and temperature limits the chips had a voltage max parameter. Precision Boost would not pick a ratio if it violated the limit. Which became a bigger deal with the second EDC "bug" where on certain AGESA if EDC set above sku default, a negative offset was applied to voltage max limit.
On Zen 3, besides the current, power, and temperature limits the chips had a voltage max parameter. Precision Boost would not pick a ratio if it violated the limit. Which became a bigger deal with the second EDC "bug" where on certain AGESA if EDC set above sku default, a negative offset was applied to voltage max limit.
would be interested to see actual results from other Iceman users to compare tbh to find out if something is wrong with mine although it just about manages to cool 450W so always assumed it was ok 😂 but yeah that Optimus block Jwick is using must be the secret sauceIt just didn't seem like from the #'s you posted (post I quoted before) that your delta is low enough. Should be 10-20C lower, at least based on what others say in regards to the temp drop from going DD. Either that or I need to buy an Optimus block.
very very impressive results with the IHS still in place and i assumed untouched
Joined
·
721 Posts
I assumed so, as mentioned here for example:
I think it clips voltages by 200 MHz, but I'm not 100% sure. The "Enhanced" version of TVB on 14th gen GPUs might also have other tricks. For example, for the 14900K Skatterbencher mentioned:
SkatterBencher #67: Intel Core i9-14900K Overclocked to 6200 MHz - SkatterBencher
We overclock and undervolt the Intel Core i9-14900K CPU up to 6.2 GHz with the GIGABYTE Z790 Aorus Master X and EK-Quantum water cooling.
skatterbencher.com
Joined
·
292 Posts
Usually 13th/14th Gen i9 by default behave:
- 2-core max boost is clipped at 70C due to TVB(how much varies per SKU)
- IABT when enabled allows the CPU to hit #8 v/f point all-core as long as the CPU is within power limits, ICCmax and temperature is lower than 70C. If temperature is hit it clips by -1 ratio until temperature remains below 70C. Max clipping is #7 v/f point which is advertised Turbo Boost 2.0 all-core frequency.
I get close with the Core1 block and right at 300l/h flow. Looks like 57C delta at 342w running R15 Extreme. Looked like Jwick had ~55C delta at ~340w.would be interested to see actual results from other Iceman users to compare tbh to find out if something is wrong with mine although it just about manages to cool 450W so always assumed it was ok 😂 but yeah that Optimus block Jwick is using must be the secret sauce
Optimus uses that insanly fine fin grid, so it may be slightly better than the Core1. Haven't seen an reviews for it though to confirm.
What is your delta at 340w?
Joined
·
2,973 Posts
don't look at package power draw as it is based on vid. POUT is always accurate.
No working for some reason. I suspect it's close though since the board sets DCLL automatically.
Joined
·
2,973 Posts
16,221 - 16,240 of 23,546 Posts
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
