Intel Core i5-13400F

[Why_Me]

Hi,
Yep vrm issues are a thing of the past, like way back to x299 days

Hi,

 

[Gica]

Hi,
Yep vrm issues are a thing of the past, like way back to x299 days

Are you sure? With all generations of processors, you will notice differences in performance between an entry level board and an overkill one.
With the 6 phases, an analogy can be made with audio amplifiers: you can build a low-cost 100W one, but it will never achieve the fidelity of a more expensive, high-end one. I have DS3H and I know what I'm talking about. It's not for nothing that they have active serial and parallel ports, because they are intended for a low cost, office system, not for intensive gaming and multitasking.

13500, Cinebench R23 multi.
Wattmeter (PEAK):
DS3H (6 phases for cores): 280W
Gaming X (16 phases for cores): 194W

Why? Because the mosfets lose their efficiency proportionally to the temperature and a motherboard with a strong VRM will use better quality components than a low entry one.
I use this DS3H with 12500+igp for activities that do not require the processor heavy multitasking: www, office, some games (WoT, AW and old), photo editing, etc.

Capture 1: 12500 + DS3H + igp, 117 minutes session.
Capture 2: Cinebench R23, 10 min stress with Gaming X

 

[W1zzard]

Just to confirm again, the 7600/7600X numbers are all wrong (well the multi-threaded numbers seem alright). I'm almost finished retesting the 7600X, have to wait for the 7600 to come back from AMD next week

 

[Lew Zealand]

And HU results always favor AMD.

I disagree. HU's opinions favor AMD but their data match what's seen elsewhere. I built my first PC based on theirs and other tests because the i5-8400 was better than the R5 2600 for the 2 things I do that were CPU-limited: gaming and h.265 conversion. The data is what matters, not their or anyone else's conclusions.

If we turn to games, tell me, how much will the 7600X increase the performance of the 3070Ti? What would be the difference between 13400/13500/7600X in this scenario? I have captured and paid attention to GPU/CPU response times.
For $7 more than 7600X, saving DDR4 memory (-$125) and at least 25% more in multithreading than the competition, I say the 13500 is a good deal. In gaming, enjoy the performance of an RTX 4090 in reviews compared to a system equipped with a weaker video card.
I don't dispute the qualities of the 7600(X), but Mr. Garrus errs when he denies the qualities of the competition.


1080p, Highest + RT Ultra

The differences with a 3070 Ti are minimal between any of those CPUs, I'm using a lower-performance CPU with a higher performance GPU but at 1440p so probably net similar. IMO the biggest difference between the 2 setups is multithread advantage of Intel CPUs like the 13400/13500 in rendering vs. the upgradability of the AM5 platform. Not everyone uses either of those advantages, makes for interesting choices.

 

[Gica]

The differences with a 3070 Ti are minimal between any of those CPUs, I'm using a lower-performance CPU with a higher performance GPU but at 1440p so probably net similar. IMO the biggest difference between the 2 setups is multithread advantage of Intel CPUs like the 13400/13500 in rendering vs. the upgradability of the AM5 platform. Not everyone uses either of those advantages, makes for interesting choices.

I think you didn't follow the discussion. 7600(X) is not a wrong choice and I don't dispute Ryzen's performance. The counters appeared when @Garrus declared that 13400(F) is a bad choice. So be it?
I chose 13500, a more expensive idea than 7600 and only 7$ than 7600X, because it offers much more than this in multithreading, especially on my target. In gaming, for my mainstream target, there is no difference between them, even with 4090 they are statistics, not visible with the eye (~5% in 1080p, ZERO in 4K). There is a small loss of performance with DDR4, but it is fully compensated by a cheaper motherboard and keeping the old RAM modules.
For future upgrades, I have available the entire 12600K to 13900KS range. There are also rumors of Raptor Refresh. There are options. For these I now purchased gaming x and not DS3H.

Gigabyte has removed from the BIOS the option to overclock non-k processors by 3%. From this moment on, Gigabyte does not exist for me. That overclock doesn't mean much, but when you steal while increasing the prices from year to year, it deserves to be banned.

I disagree. HU's opinions favor AMD but their data match what's seen elsewhere. I built my first PC based on theirs and other tests because the i5-8400 was better than the R5 2600 for the 2 things I do that were CPU-limited: gaming and h.265 conversion. The data is what matters, not their or anyone else's conclusions.

Cinebench R23 multi
HU: 20.682 pts
My: 21.200 pts

For 7zip, the discrepancy is huge
HU: 77.387
My: 98.902
From 3466 MHz to 3600 MHz is not much, but they worked with much lower latencies than mine. However, the Cinebench score does not scale with RAM speed.

And there is something else that bothers me: how come I can't reach 3600MHz with premium RAM modules and others (including me) reach this frequency with undervolt even with common RAM modules????
I set XMP and the system started from the first, no crash after 30 days of use. And 12500 had the same behavior. 3600MHz, Gear 1, works without problems with the 11th generation, but HU has problems?!?!?!

 

[Garrus]

I just got my single fan tower air cooler installed. No longer using the stock cooler.

AMD Ryzen 7600. 5.5Ghz at 1.22V.

Cinebench MT: 15,892
Cinebench ST: 1,982

 

[Gica]

13500 -> 13400
2 P-cores: 4800 -> 4600
4 P-cores: 4700 -> 4400
6 P-cores: 4500 -> 4200
E-cores: only 4 activated@3.3GHz
Same score as Tom's review, slightly higher than TPU.
Power draw: 77W (154 wattmeter, but here it differs from case to case, depending on the components that make up the entire system)
The full information looks like this (captures), plus the 15-minute endurance test (which can run for days, I don't know how stable your system is with that overclocking, nor how much it consumes)


A small test in gaming as well. The video card has a custom setting (290W -> 230W, without loss of performance). Its equivalent is the RX6800 in rasterization and the 6900XT when you activate ray tracing. I don't have a statistic, those who use equal or stronger video cards are a minority, the vast majority use weaker video cards. For all of them, 13400 is really overkill, they can manage with a powerful processor in gaming 5-6 years ago, 8700(K) as an example.

 

[THANATOS]

Really ?
You should explain to us what's wrong ...

Nice little processor, efficient, still a bit pricey.
I was expecting better things in games compared to the 12400F, but slighty better in applications ...

For a gamer point of view, it's the kind of processor i would like to see more ...
You can save some $ and invest them in a better GC, a win-win scenario

There is something I still don't get ...
Why the results are often (slighty) worse when the power limit is removed ?
Shouldn't it be the opposite ?

He already said It. The results for 7600 in the comparison 13400F vs 7600 are wrong. I already noticed this the first time I read the review.
@W1zzard please check and remake these charts, thank you.

Intel Core i5-13400F Review - Force of Efficiency

vs
AMD Ryzen 5 7600 Review - Affordable Zen 4 for the Masses

The same CPU, but a big difference in the power consumption between different reviews, but same soft or game.

 

[Garrus]

He already said It. The results for 7600 in the comparison 13400F vs 7600 are wrong. I already noticed this the first time I read the review.
@W1zzard please check and remake these charts, thank you.

Intel Core i5-13400F Review - Force of Efficiency

vs
AMD Ryzen 5 7600 Review - Affordable Zen 4 for the Masses

The same CPU, but a big difference in the power consumption between different reviews, but same soft or game.

Oh nice that makes it clear. In Cinebench in one review he had 110W. In the other 78W. etc.etc.

 

[Gica]

And at 13400 there is a small mistake: 43W games average, not 44W. A small step for efficiency, a huge one in this processor war.
If you want, even at the price of 7600X. The prices of the X series have dropped dramatically, especially the 7600X, which was at the level of 13600KF. At this moment, the price difference between 7000 and 7000X is too small to choose non-X.

Amazon.co.uk : AMD Ryzen? 5 7600X 6-Core, 12-Thread Unlocked Desktop Processor

Gigabyte's answer to the total blocking of overclocking on non-K processors on the Z chipset motherboard. Even the 3% limit blocked it.

 

[Shatun_Bear]

This should cost under $150 at $200 it's a hard sell compared to last gen processors.

 

[AleXXX666]

And HU results always favor AMD.
If we turn to games, tell me, how much will the 7600X increase the performance of the 3070Ti? What would be the difference between 13400/13500/7600X in this scenario? I have captured and paid attention to GPU/CPU response times.
For $7 more than 7600X, saving DDR4 memory (-$125) and at least 25% more in multithreading than the competition, I say the 13500 is a good deal. In gaming, enjoy the performance of an RTX 4090 in reviews compared to a system equipped with a weaker video card.
I don't dispute the qualities of the 7600(X), but Mr. Garrus errs when he denies the qualities of the competition.


1080p, Highest + RT Ultra

why on earth save retro memory when we are considering NEW platform? MAYBE one has 2133 MHz DDR4. MAY BE REALLY, because it's also "COSTS SAVING" at the time of purchase vs even 2666 or 3200 BACK THESE TIMES lol. There are very small price difference between all you "BLING BLING" 3600 CL16 DDR4 and GOOD ENOUGH 5200 CL 36 DDR5. I would DDR5 any day vs retro "COOL" DDR4. If you have B-dies which are 4400 CL 16 (I wonder if they exist but I'm not a fan of Hynix and Micron) then yes, it's stupid to buy DDR5, because you probably can't IRL all these "7000+ MHz" outside of testing lab or good OC environment lol.

 

[Gica]

Because the memory has a minor impact on performance and is not really old, DDR2 or 3 are old. Yes, I agree with the sub 3000 change.
You see, with DDR4 3600MHz you can easily set Gear 1, impossible to set with DDR5 4800+. That's why, between DDR4 3600 and DDR5 4800 you won't see any differences, they will be minimal up to 6000, and above... the price is premium.
You probably won't see anything between 3600 DDR4 and 6000+ DDR5 if the processor and video card are mainstream or weaker.

The conclusion of this review:
"That's also why I included the DDR4-3600 CL16 "sweet spot" configuration in the test results. DDR4 is readily available and has come down in pricing a lot in recent months despite the crisis and shortages. In applications, we see DDR4-3600 CL16 roughly on par with DDR5-4800, which is the lowest speed grade DDR5 announced at this time, so almost nothing is lost despite the numeric difference of "3600" vs. "4800" and huge price increase tied to it. In gaming, which is more sensitive to memory timings than memory bandwidth, our DDR4-3600 configuration works even better, achieving FPS similar to DDR5-5600. The takeaway here is to focus on DDR4 unless you absolutely want to play with the new DDR5 tech. Motherboards are cheaper, too, and you can use the DDR4 memory you already have."

 

[AleXXX666]

Are you sure? With all generations of processors, you will notice differences in performance between an entry level board and an overkill one.
With the 6 phases, an analogy can be made with audio amplifiers: you can build a low-cost 100W one, but it will never achieve the fidelity of a more expensive, high-end one. I have DS3H and I know what I'm talking about. It's not for nothing that they have active serial and parallel ports, because they are intended for a low cost, office system, not for intensive gaming and multitasking.

13500, Cinebench R23 multi.
Wattmeter (PEAK):
DS3H (6 phases for cores): 280W
Gaming X (16 phases for cores): 194W

Why? Because the mosfets lose their efficiency proportionally to the temperature and a motherboard with a strong VRM will use better quality components than a low entry one.
I use this DS3H with 12500+igp for activities that do not require the processor heavy multitasking: www, office, some games (WoT, AW and old), photo editing, etc.

Capture 1: 12500 + DS3H + igp, 117 minutes session.
Capture 2: Cinebench R23, 10 min stress with Gaming X

THANK YOU very much for info about phases. I have B660M DS3H with 12100 and would like to upgrade CPU either to 12400F or 13400F. I don't have either budget or plan to change the MB, so I would choose 12400F!

Because the memory has a minor impact on performance and is not really old, DDR2 or 3 are old. Yes, I agree with the sub 3000 change.
You see, with DDR4 3600MHz you can easily set Gear 1, impossible to set with DDR5 4800+. That's why, between DDR4 3600 and DDR5 4800 you won't see any differences, they will be minimal up to 6000, and above... the price is premium.
You probably won't see anything between 3600 DDR4 and 6000+ DDR5 if the processor and video card are mainstream or weaker.

The conclusion of this review:
"That's also why I included the DDR4-3600 CL16 "sweet spot" configuration in the test results. DDR4 is readily available and has come down in pricing a lot in recent months despite the crisis and shortages. In applications, we see DDR4-3600 CL16 roughly on par with DDR5-4800, which is the lowest speed grade DDR5 announced at this time, so almost nothing is lost despite the numeric difference of "3600" vs. "4800" and huge price increase tied to it. In gaming, which is more sensitive to memory timings than memory bandwidth, our DDR4-3600 configuration works even better, achieving FPS similar to DDR5-5600. The takeaway here is to focus on DDR4 unless you absolutely want to play with the new DDR5 tech. Motherboards are cheaper, too, and you can use the DDR4 memory you already have."

ok, so 3600 CL 16 is better than 3733 CL 19? I would like to upgrade RAM and some info there is that bandwidth is also important and I'd like to choose RAM with maximal bandwidth available with selected ram size (2x16 GB) but it should be in my mobo QVL list so I won't do idiot mistake again like buying 4400 ram and not possible to run XMP lmfao

 

[sLowEnd]

THANK YOU very much for info about phases. I have B660M DS3H with 12100 and would like to upgrade CPU either to 12400F or 13400F. I don't have either budget or plan to change the MB, so I would choose 12400F!

Your motherboard would be fine with either chip.

Gica appears to subscribe to the notion that more is always better when it comes to VRMs, when that isn't really the case. Overkill is not always tangibly better.

Modern VRMs are stupid - Cultists Network

As motherboard VRMs tend to go more and more overkill every generation, they are also becoming more and more stupid.

cultists.network

 

[Gica]

THANK YOU very much for info about phases. I have B660M DS3H with 12100 and would like to upgrade CPU either to 12400F or 13400F. I don't have either budget or plan to change the MB, so I would choose 12400F!


ok, so 3600 CL 16 is better than 3733 CL 19? I would like to upgrade RAM and some info there is that bandwidth is also important and I'd like to choose RAM with maximal bandwidth available with selected ram size (2x16 GB) but it should be in my mobo QVL list so I won't do idiot mistake again like buying 4400 ram and not possible to run XMP lmfao

Your motherboard also handles well with 13400F. With 13500 or above, it's a problem with renderings, encodings and any application that takes the processor consumption over 100W for a long time. There are no problems with the games.

Above 3600MHz, there is the problem of stability with GEAR 1. Officially, the limit is 3200, but all processors can work in GEAR 1 with memories clocked at 3600. Higher speeds with GEAR 2 are not worth it because the memory controller works at 1/2 speed.

 

[W1zzard]

Finally! The review has been updated with corrected numbers for the 7600 and 7600X. I've also retested the other Zen 4 CPUs on the 9922 BIOS with the newest chipset drivers.

While Ryzen 7600/7600X have gained quite a bit in gaming performance and power, the outcome of the conclusion remains unchanged

 

[Wasteland]

Officially, the limit is 3200, but all processors can work in GEAR 1 with memories clocked at 3600

Not true. Owners of locked (non-K) Alder/Raptor Lake parts should expect no better than 3466 MT/s stable in Gear 1. If they want to save a little money and potentially a lot of time, they're probably best off just buying 3200 MT/s memory, hitting the XMP button, and moving on with their lives. Recommending higher memory speeds in Gear 1 is irresponsible. Sure, it's possible to get a golden sample, but that possibility is overwhelmed by the larger chance that people who have neither knowledge of or interest in memory overclocking (or more to the point, stability testing) will screw themselves.

(Really at this point new buyers are probably best off with DDR5, but the argument stands re DDR4.)

I know you claimed earlier that this memory speed limitation for locked Intel parts was a conspiracy theory on the part of Hardware Unboxed, because you think they hate Intel, but HUB didn't even talk about it until a few months ago; in fact they recommended locked Alder parts for quite some time with nary a mention of System Agent Voltage--probably because at that time, HUB did all its testing with 3200 MT/s B-die, and so the issue never came up. This oversight on HUB's part actually annoyed me, because I had personal experience with memory instability in Gear 1 with two separate locked Intel CPUs. I wrote at least one lengthy comment criticizing HUB for their oversight, and I was not alone. Buildzoid, to his credit, got it right early on--not a surprise given that memory overclocking is pretty much the singular focus of his channel--and it's likewise to HUB's credit that they're now collaborating with Buildzoid on memory tuning.

Meanwhile, MSI released BIOS updates to force Gear 2 at 3600 MT/s, because the alternative was too unstable. Techpowerup's forums, too, were talking about it. It's one of those issues that's really only annoying precisely because it isn't prominently discussed. If you google anything to do with Intel CPUs, most of the time you'll get a million results about K-SKUs. It didn't help, in this case, that Intel went out of its way to tout the memory overclocking ability on its B-series motherboards for LGA 1700--traditionally not overclocking boards. There are probably a lot people out there running in Gear 2 (or worse, running unstable) without realizing it, because they read a bunch of articles claiming that 3600 MT/s memory is the "sweet spot," based on testing with e.g. a 12900k.

Like I say, this issue isn't a big deal. It's just something that might annoy or inconvenience the user who doesn't know about it ahead of time. At worst, it could lead to hidden instabilities that last for weeks or months, and possibly data corruption as a result.

 

[Gica]

The voltage that supplies the memory controller is called VDDQ TX voltage. SA voltage only helps, probably in case of extreme overclocking.
VDDQ TX voltage:
1.2V default (no XMP)
1.3V XMP ON
1.1V for me, with the memories set to 3600MHz and the memory controller in GEAR 1, four slots occupied.
I also have a 12500 and it behaves identically. Of course, I use it in another system, with DDR4 3200 and GEAR 1, but it was used for a long time with the memories from the capture and it works without any problems.
I don't know where the irresponsibility is with the undervolt from 1.3V to 1.1V???? Even at 1.3V there are no problems, this value being established by Intel.
SA is locked at 0.9V. In the 6-10th series (based on skylake) it would have an important impact for memory overclocking, but this is no longer the case with Alder and Raptor.
Neither HU nor anything else teaches me anything extra about the processors I own.

 

[Wasteland]

No one cares what you do with your CPUs, which is why I didn't reply until you recommended to someone else that he should expect 3600 MT/s in Gear 1. That's the irresponsible part.

Posting screengrabs of HWINFO doesn't prove stability, and those screengrabs are a pretty lame rebuttal to all of the links I posted. If your system is stable, then congrats, more power to you, enjoy, etc. Most people are better off avoiding the headache.

 

[Gica]

No one cares what you do with your CPUs, which is why I didn't reply until you recommended to someone else that he should expect 3600 MT/s in Gear 1. That's the irresponsible part.

Posting screengrabs of HWINFO doesn't prove stability, and those screengrabs are a pretty lame rebuttal to all of the links I posted. If your system is stable, then congrats, more power to you, enjoy, etc. Most people are better off avoiding the headache.

It seems irresponsible to me when you make statements in ignorance of the cause. You are trying to transfer the software problems to the processor, my god!!! All Socket 1700 motherboards have at least one BIOS update that increases RAM compatibility or stability. All, from the most entry model!

I took advantage of installing a game to show you what I know since I bought the processor. Do you see any signs of instability? In parallel with the game (and youtube & news) I also started the AIDA utility for stress. Almost 2 hours session, 51 minutes stress. I am also attaching a video file because I expect from you even accusations of rigging.
Two questions:
1. If 3600 GEAR 1 (1800MHz controller) is dangerous, how about 7200 or more GEAR 2 (1800+ MHz controller)?
1. What is the difference between a stock processor with 1.25V and one with oc +200MHz and undervolt 0.2V?

Stop making accusations where you don't understand!

 

[Wasteland]

It seems irresponsible to me when you make statements in ignorance of the cause. You are trying to transfer the software problems to the processor, my god!!! All Socket 1700 motherboards have at least one BIOS update that increases RAM compatibility or stability. All, from the most entry model!

I took advantage of installing a game to show you what I know since I bought the processor. Do you see any signs of instability? In parallel with the game (and youtube & news) I also started the AIDA utility for stress. Almost 2 hours session, 51 minutes stress. I am also attaching a video file because I expect from you even accusations of rigging.
Two questions:
1. If 3600 GEAR 1 (1800MHz controller) is dangerous, how about 7200 or more GEAR 2 (1800+ MHz controller)?
1. What is the difference between a stock processor with 1.25V and one with oc +200MHz and undervolt 0.2V?

Stop making accusations where you don't understand!

LMAO. Do you know how those BIOS updates "increase memory stability?" By defaulting to Gear 2 when the user dials in XMP speeds of 3600 MT/s or higher. If Alexx or whoever wants to believe you over HUB, Buildzoid, MSI, and frankly Intel itself at this point, then they're certainly free to do so--but you're not winning over anyone sane with this argument. Posting over and over again, "But I got it working, therefore VCCSA is irrelevant" does not make the case.

It looks like you got a good sample, though what you've shown isn't exactly ironclad proof of stability. (It is, of course, impossible to prove a negative--the absence of instability--which is why very heavy stress testing is recommended for long-term overclocks.) First thing I'd do in your shoes is run stressapptest for about one hour per 16 GBs of RAM; I do it on all my machines now whether I'm overclocking the memory or not. It's easy and fast, relative to most alternatives. But like I say, no one really cares what you do or don't do with your system. We're arguing about the general case, and in that context you have no leg to stand on.

 

[Gica]

The stability test is over 30 days since I use it. I don't allow system crashes in online games. A good word can put 12500, but it can't. It now runs with other memories at 3200MHz because that's what xmp has.
Unsafe is that situation where you have to increase the voltage above the specifications to maintain stability at a certain frequency.

I don't want to prove anything. Apart from HU, all reputable reviewers have tested Alder and Raptor at 3600 without problems. Keep clinging to this nonsense with "unsafe".

By the way, how many 7700K owners went "safe" with memories clocked up to 2400 MHz? How many 6700K owners have considered "unsafe" memory clocking above 2133 MHz? Let's continue?

 

[AleXXX666]

Your motherboard also handles well with 13400F. With 13500 or above, it's a problem with renderings, encodings and any application that takes the processor consumption over 100W for a long time. There are no problems with the games.

Above 3600MHz, there is the problem of stability with GEAR 1. Officially, the limit is 3200, but all processors can work in GEAR 1 with memories clocked at 3600. Higher speeds with GEAR 2 are not worth it because the memory controller works at 1/2 speed.

View attachment 288646

ok. i chose 12400F because no-brain price-to-performance ratio. 13400F is much pricier but gives less % in performance compared to price % increase lol.
then i chose 3733.19 kingston fury beast ram which has second profile 3600.17. i could chose fury renegade which are 3600.16 but i preferred beast because i hate "tall" ram lol. Maybe I'll change my cooler again (rn i have hyper 212) and so I won't mess up and choose among only "RAM friendly" coolers but instead select something which isn't overhyped & overpriced also I searched for RAM which is available in my region and in QVL with my Giga B660M DS3H as I decided to not change mb.

Your motherboard would be fine with either chip.

Gica appears to subscribe to the notion that more is always better when it comes to VRMs, when that isn't really the case. Overkill is not always tangibly better.

Modern VRMs are stupid - Cultists Network

As motherboard VRMs tend to go more and more overkill every generation, they are also becoming more and more stupid.

cultists.network

yeah I though so but 12400F is a no-brain choice over next gen in terms of price to performance.

 

[Gica]

Gica appears to subscribe to the notion that more is always better when it comes to VRMs, when that isn't really the case. Overkill is not always tangibly better.

Not. For gaming, DS3H it's all you need at a good price. For powerful processors used in intensive tasks, with consumption over 100W, it is not the best choice. I tested 13500 and the VRM temperature reaches 100 degrees only in the simple run of the Cinebench R23 benchmark. With the sensor indicating 100 degrees, there are at least 120 degrees inside the mosfets and stability problems are guaranteed. This is not the case with the 12500 (I have been using it with this motherboard since June 2022), nor with the 13400. In addition, a motherboard with a better VRM provides you with the reserve for a CPU upgrade in the future.
I also disagree with the 12500 versus 13500 comparison. The 13500 beats the 12500 by more than 50% in multitasking and costs only maximum 20% more. With us, the difference is only $30.
It's a little beast at a good price.