*Official Benchmark Thread* - Post it here or it didn't happen :D

[Talon]

 

Higher performance with APO, lower power consumption. 

 

Intel being a bunch of di**s not giving this to 12th and 13th gen though. I am happy to see HWU shit all over Intel in this case because they deserve it. APO, if implemented into enough games moving forward should be on all of their LGA 1700 hybrid CPUs. This type of anti consumer behavior is BS and needs to be called out as such. 

[Mr. Fox]

18 minutes ago, Talon said:

 

Higher performance with APO, lower power consumption. 

 

Intel being a bunch of di**s not giving this to 12th and 13th gen though. I am happy to see HWU shit all over Intel in this case because they deserve it. APO, if implemented into enough games moving forward should be on all of their LGA 1700 hybrid CPUs. This type of anti consumer behavior is BS and needs to be called out as such. 

That is a very "NVIDIA" move on Intel's part.

[tps3443]

1 hour ago, Talon said:

 

Higher performance with APO, lower power consumption. 

 

Intel being a bunch of di**s not giving this to 12th and 13th gen though. I am happy to see HWU shit all over Intel in this case because they deserve it. APO, if implemented into enough games moving forward should be on all of their LGA 1700 hybrid CPUs. This type of anti consumer behavior is BS and needs to be called out as such. 

Yeah that is pretty annoying. I’m just glad the gains are fairly minor. Not nearly as bad as me missing out on the Frame Gen feature that the RTX40’s have. This one feature alone would make it worth getting even just a RTX4070Ti. Lower power and high frames! Ark Ascended is really hard to run at 4K with no Frame Gen.

 

@Mr. Fox Did you try those Team Groups in the Apex Z790 yet? My G.Skill 7200’s can run 8600C38 solid with 1.64V VDD, and 1.590V VDDQ. Latency is about 51ns.
 

I have backed off the CPU IMC voltage, and backed off the IVR TX voltage to 1.400V or less range, and then I started juicing up the DDR5 VDD/VDDQ and I am seeing some really really good results now with bios 1501. 

 

[Mr. Fox]

28 minutes ago, tps3443 said:

@Mr. Fox Did you try those Team Groups in the Apex Z790 yet? My G.Skill 7200’s can run 8600C38 solid with 1.64V VDD, and 1.590V VDDQ. Latency is about 51ns.

No, I have decided that it was very silly for me to spend money on something that technically isn't an upgrade. I dont need the extra 16GB so I am not going to burn any more calories tuning something I have no real use for. These sticks seem to work pretty much the same as the G.SKILL and I don't want to mess with the Apex. It's working as I want it to work and if I start messing with it that could change. I think I am going to use the money for something actually useful instead of the memory kit that I don't need.

[electrosoft]

18 hours ago, Mr. Fox said:

I hope the latter is true. If they want to be taken as seriously as EVGA was by enthusisasts, they'll need to not waste a lot of time and energy with mainstream consumer garbage, at least in the enthusiast space products if they plan to dominate there. And, they'll need to provide a warranty as good as EVGA provided for those products. The three to 6+ weeks turn-around time on warranty service, with outbound shipping paid by the victim, as we have come to expect from ASUS, is totally unacceptable. Taking a wait-and-see approach and only building/selling what is popular and sells well, as smart as it might be, is not very likely. It would be difficult to find fault with it, though.

 

Unlike EVGA, MSI has the size to double down on a true enthusiast board and GPU while also continuing to cater to other segments of the market with the glitz, RGB and aesthetic approach.

 

If Vince and his team (or whoever he adds/subtracts) gets in there and focuses on a KPE 4090/5090 GPU and a KPE Z790/Z890 Unify level product targeting enthusiasts and overclockers and that's it, I'm all in for that.

 

While it looks like MSI has retreated from the AMD GPU market (without saying anything), their AM5 motherboard offerings are massive with 15 different models from A620 (1), B650 (9),  X670 (1) and X670e (4) so they may be becoming Nvidia exclusive but still offering both Intel and AMD motherboards.

 

10 hours ago, Mr. Fox said:

 

Well, not a bad kit. Quality is fine. The advertisement (even on the box) of a "one piece design" on the heatsink is just straight-up false. It is two pieces. The seam is not obvious at a glance and it looks like one piece. Not sure I am going to keep it because it is not remarkably  better than the RAM I took out of the Dark. It runs the same memory overclock as the G.SKILL kit on the Apex. I will think about it for a day or two. Not really gaining anything for the money spent. The 16GB*2 generic green A-die sticks are basically just as good. The only measurable benefit, which may not even matter in the grand scheme of things, is the extra 16GB of memory capacity. The extra capacity is worth something, but I am not sure that is a good enough basis for keeping versus RMA/refund.

 

Here is the XMP default.

Here is the memory overclock matching the Apex.

If I keep it then, of course, it will have the heating blankets removed and the water-cooling kit on the generic A-die sticks will be installed on them. Just a matter of whether or not I decide it is worth the extra money and associated hassle. I'm kind of leaning toward "no" but we will see if I change my mind in a few days. I gained nothing on overclocking capacity, but also didn't lose anything except the slightly tighter timings possible with the 16GB first-generation A-die modules.

 

Is there a chance you're bouncing up against a motherboard/CPU limitation yet?

[Mr. Fox]

8 hours ago, electrosoft said:

Is there a chance you're bouncing up against a motherboard/CPU limitation yet?

With the 24GB modules, EVGA released one BIOS update to accommodate them before "getting out of the business" and letting their firmware team go. I doubt there will be any more firmware updates and I think I might be seeing the firmware not being optimized for the new 24GB modules. So, no on the CPU and motherboard, but probably yes on the firmware. The 13900KS in the Dark came out of the Apex and it has a stronger IMC than the one in the Apex. The 13900K (sitting on a shelf) is not as good on the P and E core bin, but the IMC on that CPU is stronger than both of the 13900KS. CPU overclocking on all three CPUs is very noticably better on the Dark using the 16GB A-die than it is on the Apex, but the memory overclocking is better on the Apex with the 24GB A-die modules. The 16GB A-die modules overclock better on the Dark than they do on the Apex. It is difficult to get them stable on the Apex using recent firmware.

 

This is the older A-die (16GB generic modules). Doing this with the newer 24GB A-die modules (G.SKILL or TG) is more difficult and finicky with the newer modules, and I believe that is due to firmware not being optimized for them. The opposite is true for the Apex. It is optimized for the 24GB modules and not the 16GB. It takes less to upset the apple cart with the 16GB modules and the timings can be run much tighter. As a result of the tighter timings, the read/write/copy speeds and latency are better with the 16GB modules at a couple hundred MHz less compared to the Apex.

 

 

[electrosoft]

APO explained / theorized (E cores doing much more work, isolating L2/3 cache workloads, less ring work)....

 

And Intel's official response to supporting 12th and 13th gen:

 

"Intel has no plans to support prior generation products with Application Optimization"

 

What an absolute garbage response but seeing as it only supports a few games and at 4k the results are nearly a wash it's ok but still what a craptastic Nvidia approach to something that is software based not being implemented on previous generations especially 13th which is basically identical.

 

Plus their response to additional titles/selections adding APO support:

 

"Intel will continue to evaluate additional applications and games for support on an ongoing basis, this will include select current and future triple A game titles"

 

In other words, you're at the mercy of Intel and what titles they optimize for APO support.

 

 

Like @Talon said, d*ck move.

 

You are also left to wonder why these titles were selected (probably showcased it in the best light) and how many titles bring nothing to the table as we have seen can happen with AMD's X3D from time to time.

[tps3443]

This is awesome. 
 

 

[electrosoft]

4090's are drying up like crazy everywhere in the US:

 

Newegg is completely sold out of "shipped and sold by Newegg" 4090s even with the substantial price hikes.....

Best Buy is still 100% out of stock

My closest local Microcenter is now sold out (St Davids)

Second closest Parkville MC has 8 total left in stock (5 Suprim, 3 Asus AIO)

Third Closest Brooklyn, NY is now completely sold out

Lowest priced 4090 on Amazon is >$2000 with three models in stock (TUF $2040, Windforce $2100, White Asus Strix $2850) and all third party sellers at this point

 

 

Still plenty of 7900xtx's or 4080's @Papusan  EVERYWHERE for you to add to your collection! 🤣

 

 

 

 

 

[Papusan]

On 11/5/2023 at 9:04 PM, Papusan said:

I also hate the stupidity with onboard gen_5 M2 slot above the GPU. Why implement this on a OC board? Forced by Intel? Not so sure about that. Why not offer the same old PCIe 5.0 box for the M.2 drive as for first gen Apex 1700 boards? Those that want gen 5 drives could use that black PCIe box instead of Asus infecting this board with an useless feature.

Yup. See... No PCIE Gen5 M.2 Slot is needed. This means the stupids at Asus implemented an useless feature for their only OC bords. Who want castrate some of the PCIe lines for their graphics cards for a feature you almost never will use on gaming and OC boards? Because Intel don't force you to do this stupidity. Shame on you... Asus. @Mr. Fox Whats wrong with some of todays engineers? 

 

3 hours ago, electrosoft said:

Still plenty of 7900xtx's or 4080's @Papusan  EVERYWHERE for you to add to your collection! 🤣

 

Maybe I will buy a 4080 when the prices drop to $50-80 USD. But I will most likely never spend my money on a used AMD card. Has to be damn cheap. More like $30-40 bucks 🙂

 

 

Edit. If I was't banned from Cablemod reddit I would say he was a liar....  Sad that they prefer ban everyone that tell the truth. 

 

 

[Rage Set]

1 hour ago, electrosoft said:

APO explained / theorized (E cores doing much more work, isolating L2/3 cache workloads, less ring work)....

 

And Intel's official response to supporting 12th and 13th gen:

 

"Intel has no plans to support prior generation products with Application Optimization"

 

What an absolute garbage response but seeing as it only supports a few games and at 4k the results are nearly a wash it's ok but still what a craptastic Nvidia approach to something that is software based not being implemented on previous generations especially 13th which is basically identical.

 

Plus their response to additional titles/selections adding APO support:

 

"Intel will continue to evaluate additional applications and games for support on an ongoing basis, this will include select current and future triple A game titles"

 

In other words, you're at the mercy of Intel and what titles they optimize for APO support.

 

 

Like @Talon said, d*ck move.

 

You are also left to wonder why these titles were selected (probably showcased it in the best light) and how many titles bring nothing to the table as we have seen can happen with AMD's X3D from time to time.

 

All of the major companies are taking a piss on consumers. AMD with how they handled the Starfield situtation and now with the Vega drivers. Intel with APO and continued product segmentation. Finally Nvidia with ... everything 40 series. If all of this doesn't make fanboys understand that none of these companies are your friends, I don't know what will.

[Talon]

1 hour ago, Rage Set said:

 

All of the major companies are taking a piss on consumers. AMD with how they handled the Starfield situtation and now with the Vega drivers. Intel with APO and continued product segmentation. Finally Nvidia with ... everything 40 series. If all of this doesn't make fanboys understand that none of these companies are your friends, I don't know what will.

 

💯

[Mr. Fox]

1 hour ago, Papusan said:

Shame on you... Asus. @Mr. Fox Whats wrong with some of todays engineers? 

There are many things I have always believed, and with a few things it is less of a belief now and more of an observation of fact. A belief is something you expect to be true that may not be supported by tangible evidence. It may be 100% true and accurate, but you cannot always produce tangible evidence that establishes it as being an indisputable fact that remains in the absence of belief or acceptance. Two things, in particular, I no longer need to believe because the evidence dispels any element of doubt or uncertainty. The human race is characterized by two natural traits that are only overcome by a desire or compulsion to do what is unnatural. Human beings are born evil and stupid, and will perfect both of those character flaws to the point of severity unless they are trained to be the exact opposite. The answer to your question is "because they are evil and stupid" and the surplus of stupidity among those accepting of them yields the outcome they seek.

1 hour ago, Rage Set said:

 

All of the major companies are taking a piss on consumers. AMD with how they handled the Starfield situtation and now with the Vega drivers. Intel with APO and continued product segmentation. Finally Nvidia with ... everything 40 series. If all of this doesn't make fanboys understand that none of these companies are your friends, I don't know what will.

Amen to that. And, very much in alignment with the above. It does not matter if anyone disagrees or does not believe it is true because the evidence demands a verdict in spite of anything a fanboy may want to believe to the contrary. Optimism in what they want to believe when faced with facts to the contrary highlights their abject and unforgivable stupidity. It is not open to debate and it cannot be fixed. The only thing left to decide is how long you are going to tolerate it before you drive a wooden stake through its chest.

[tps3443]

This is a pretty speedy ram profile with some fairly loose timings. I have been having some strange things happen with ram OC lately since new bios and it was from pushing too much voltage to my IMC and IVR TX, since starting from scratch on bios 1501, I did NOT adjust CPU IMC voltage or IVR TX voltage at all. I left both on auto and miraculously DDR5 8600 C38 works very well. Latency on @Mr. Fox benching windows OS would probably be like 50NS. 

Ram voltages in bios are 1.640V VDD, and 1.590V VDDQ.

[Mr. Fox]

42 minutes ago, tps3443 said:

This is a pretty speedy ram profile with some fairly loose timings. I have been having some strange things happen with ram OC lately since new bios and it was from pushing too much voltage to my IMC and IVR TX, since starting from scratch on bios 1501, I did NOT adjust CPU IMC voltage or IVR TX voltage at all. I left both on auto and miraculously DDR5 8600 C38 works very well. Latency on @Mr. Fox benching windows OS would probably be like 50NS. 

Ram voltages in bios are 1.640V VDD, and 1.590V VDDQ.

Indeed. Leaving the VDDQ TX voltage on auto seems to increase latency. I totally forgot about that. I can run it like 1.275V on the Dark, but if I bump it to 1.400V my latency drops to ~50ns and the L2 and L3 cache speeds increase. I keep forgetting this for some reason. I discovered it months ago when you made the comment to me about how low it was (before I purchased the Z790 Apex).

 

As you can see, at only 8200 this is within strking distance to 8600 on the Apex and look how high the write and copy speeds are, with 50.3 ns latency. And, that is less than 1.500V memory voltage. The 48GB 8000 kits won't boot with timings this tight. You have to loosen up a lot of them with the 48GB A-die.

 

[tps3443]

22 minutes ago, Mr. Fox said:

Indeed. Leaving the VDDQ TX voltage on auto seems to increase latency. I totally forgot about that. I can run it like 1.275V on the Dark, but if I bump it to 1.400V my latency drops to ~50ns and the L2 and L3 cache speeds increase. I keep forgetting this for some reason. I discovered it months ago when you made the comment to me about how low it was (before I purchased the Z790 Apex).

 

As you can see, at only 8200 this is within strking distance to 8600 on the Apex and look how high the write and copy speeds are, with 50.3 ns latency. And, that is less than 1.500V memory voltage. The 48GB 8000 kits won't boot with timings this tight. You have to loosen up a lot of them with the 48GB A-die.

 

Yes you’re right those 2x16’s work well on the Dark and that’s probably what you should leave in it, but nice 2x24’s can absolutely smash in an Apex. My timings are crap and loose. My memory is absolutely below average for water chilled ram. Even worse than your 8000 kit, but it’s all that I can get stable for 8600c38. If you go key in Sugi’s timings for 8600 just for benching purposes with 2x24GB it’s like 45NS with maybe 1.450VDD which is insane, I can run that for about all of about 3-6 minutes before a BSOD occurs haha. But he is using really good Hynix green sticks 2x24GB kinda like your green stick 2x16’s in that dark. 
 

If I was smart I would have bought some binned rams on HWbot when they pop up from time to time, and I’d be running a fully tuned 8600 profile. 

[Mr. Fox]

10 minutes ago, tps3443 said:

Yes you’re right those 2x16’s work well on the Dark and that’s probably what you should leave in it, but nice 2x24’s can absolutely smash in an Apex. My timings are crap and loose. My memory is absolutely below average for water chilled ram. Even worse than your 8000 kit, but it’s all that I can get stable for 8600c38. If you go key in Sugi’s timings for 8600 just for benching purposes with 2x24GB it’s like 45NS with maybe 1.450VDD which is insane, I can run that for about all of about 3-6 minutes before a BSOD occurs haha. But he is using really good Hynix green sticks 2x24GB kinda like your green stick 2x16’s in that dark. 
 

If I was smart I would have bought some binned rams on HWbot when they pop up from time to time, and I’d be running a fully tuned 8600 profile. 

I may give this 48GB kit one more go before I send it back. I have noticed it is more difficult to get my memory OC stable than it used to be and I don't know if it is the IMC on the 13900KS is that much weaker than the 13900K or if I am wearing out the generic A-die modules from overclocking the crap out of them for so long. These memory modules have really been run through the wringer during my overclocked benching tirades. They may be fatigued, which is why I decided to try one of these 48GB kits. If I do send it back, I may try dropping the 13900K back into the Dark to rule out the weaker IMC.

 

I won't be able to tell for certain with the 48GB kit running on air and if I pull off the heating blankets and put them on water I have to be committed to keeping them even if I don't want to. I cannot return them in good conscience having done that to them. It would be wrong for the person that buys them as an open box return. I know I would be really upset if I was that customer.

[ryan]

geforce now is the future.

[tps3443]

6 minutes ago, ryan said:

geforce now is the future.

How’s that 4080 running?! Lol. I know GeForce now is really a cool idea, and great service. But how is the image quality can you tell it’s streamed? I know my Gamepass Cloud gaming looks like crap compared to real life. It’s like hazy and fuzzy looking and it almost looks like a YouTube video of gameplay compared to real gameplay. But GeForce now may be different. Never tried it. 

[Mr. Fox]

33 minutes ago, ryan said:

geforce now is the future.

Zombie gaming at its best, LOL. 

25 minutes ago, tps3443 said:

How’s that 4080 running?! Lol. I know GeForce now is really a cool idea, and great service. But how is the image quality can you tell it’s streamed? I know my Gamepass Cloud gaming looks like crap compared to real life. It’s like hazy and fuzzy looking and it almost looks like a YouTube video of gameplay compared to real gameplay. But GeForce now may be different. Never tried it. 

If the screenshot is representative of that virtual 4080's image quality I'm thinking not very good. But, I have also no experience with it.. Nor do I want to. I don't like the idea of "subscribing" to most things. I prefer to buy once, own it forever and never spend another dime on that software product.

[ryan]

we the consumer are being forced to own nothing. linux is all that will be left and I guess thats not a bad thing

 

as for the image quality, right out of the box it looks awful. tweaking it a bit and sitting 10 feet away with the ai upscaling it can look strikingly good. I don't know I posted pics of 4k geforce now ultimate. Im probably not going to buy a desktop once I get my internet 1gbps. It will probably look native with the 1 gig internet, but if you want to play CS 2 or seige ect good luck as lag is present. its perfect for a handful of titles such as third person adventure games like starfield and racing games like motorfest. but for games like cod its just not there yet. I also see geforce now as a method for big uncle sam to create a algorithm to know your every move in life by collecting all the data and input you do with games. theres always a bigger picture and clandestine motives with big companies.

 

 

thats shot with crysis remastered, its the only game that doesn't look native it has compression artifacts with my 100mbps internet im sure once i get the boost in a few days it will improve but by how much is yet to be seen, its definitely a 4080 and its not castrated. but yea 4k/60 for video using geforce now is going to need a zippy internet connection. I think 250mbs and up is needed even though they of course say 75mps is fine. but what you get looks awful sometimes with the compression artifacts, it looks clunky. but yeah from 10 feet back and on motorfest it was crystal sharp.

[anassa]

Found an old P150EM Timespy benchmark:

 

3920xm and 1070

 

 

Cinebench R15 for 3920xm

 

 

Same setup with firestrike:

 

 

Good old days!

 

I wonder how the Quadro RTX 3000 would work!   

[Mr. Fox]

8 hours ago, anassa said:

Found an old P150EM Timespy benchmark:

 

3920xm and 1070

 

 

Cinebench R15 for 3920xm

 

 

Same setup with firestrike:

 

 

Good old days!

 

I wonder how the Quadro RTX 3000 would work!   

Indeed, those where the days, and my... how things have changed. Not for the better, especially in turdbook land. RTX 3000 would likely work fine, but may need a BIOS update.

 

23 hours ago, Talon said:

 

Higher performance with APO, lower power consumption. 

 

Intel being a bunch of di**s not giving this to 12th and 13th gen though. I am happy to see HWU shit all over Intel in this case because they deserve it. APO, if implemented into enough games moving forward should be on all of their LGA 1700 hybrid CPUs. This type of anti consumer behavior is BS and needs to be called out as such. 

I have been a Park Control and Process Lasso user for years... long before the stupid ideas of compromised smartphone chiplets and tablet Atom cores began to spoil our desktop fun. Both utilities work wonders on glorious high core count monolithic processors and core utilization has never been an issue for me using versions of W7 and W10 that allegedly could not function correctly without a downgrade to W11 cancer to manage the "scheduling" of them. I am wondering if this is not the better way anyhow. Use 'em all... all the time. No reason not to. Maybe having the ability to predetermine core affinity and the assign behavior one desires is just too much effort for today's push-button gamers and lazy AI-overclocking children.

[Papusan]

On 11/14/2023 at 2:18 AM, Rage Set said:

All of the major companies are taking a piss on consumers. AMD with how they handled the Starfield situtation and now with the Vega drivers.

Who can they blame? On the consumers or themselves? Todays word has to be Greed.

 

 

AMD know this was fake numbers. But they needed the first months on top to try sell their cards.... I wonder how bad the sales of AMD cards have been if Bethesda wasn't paid by AMD to screw nvidia and Intel. 

 

AMD is the exclusive PC partner for Starfield amd.com

Plays best on AMD Radeon RX Graphics..........

 

Starfield's new patch finally gives NVIDIA GPU owners the frame rates they'd expect

https://www.tweaktown.com/news/94329/starfields-new-patch-finally-gives-nvidia-gpu-owners-the-frame-rates-theyd-expect/index.html

 

--------------------------------------------------------------------------------

 

 

The RTX 4090 Matrix for more than $8000 in China! overclocking.com

[electrosoft]

38 minutes ago, Papusan said:

Who can they blame? On the consumers or themselves? Todays word has to be Greed.

 

 

AMD know this was fake numbers. But they needed the first months on top to try sell their cards.... I wonder how bad the sales of AMD cards have been if Bethesda wasn't paid by AMD to screw nvidia and Intel. 

 

AMD is the exclusive PC partner for Starfield amd.com

Plays best on AMD Radeon RX Graphics..........

 

Starfield's new patch finally gives NVIDIA GPU owners the frame rates they'd expect

https://www.tweaktown.com/news/94329/starfields-new-patch-finally-gives-nvidia-gpu-owners-the-frame-rates-theyd-expect/index.html

 

--------------------------------------------------------------------------------

 

 

The RTX 4090 Matrix for more than $8000 in China! overclocking.com

 

I'm sure AMD exclusivity included some official and unofficial conditions for the launch window and free game bundle with their GPUs and CPUs. Officially optimized for AMD. Unofficially? Well...... 🤣

 

Crazy performance gains at NATIVE no upscaling on either side.

 

13900k/14900k + 4090 is clearly going to be the optimal combo for this game now.

 

Nvidia went from being behind 8% to jumping ahead 23% with the 4090 over the 7900xtx which is more in line with real world expected results (if not even higher).

 

I'm glad I held off playing till patches galore.

 

@tps3443 you'll need to get back in there with your KPE 3090 and see how much better it is now with the update.