Tex

4.1K posts

Tex

Tex

@TEX_cs

#SiliconGang #superconductive_society Intel fanboy brainlet.

India Sumali Ekim 2016
65 Sinusundan62 Mga Tagasunod
Tex
Tex@TEX_cs·
@coolingreviews They forgot to enable the "default performance profile". /s
English
0
0
2
15
Albert Thomas
Albert Thomas@coolingreviews·
Are you telling me that the F-35 is like an overclocked and unstable CPU?
Albert Thomas tweet media
English
3
0
4
160
Jaykihn
Jaykihn@jaykihn0·
Wildcat Lake SKUs and clocking
Jaykihn tweet media
English
5
6
64
6.6K
Tex
Tex@TEX_cs·
@DreadyBear @lackynamber1234 ARL-R should have been a smaller 6+8 cpu with 48 or 64mb L3 cache. The die size would have been smaller than 90mm2, gaming performance closer to 9800X3D than 7800X3D, while surpassing both in MT.
English
0
0
3
88
DreadyBear
DreadyBear@DreadyBear·
@lackynamber1234 My experience with Arrow Lake was fantastic, and to me just extrapolating what a Nova Lake bLLC would look like from Arrow Lake vs vanilla Zen 5, I think there could be an upset if Intel doesn’t fart around. Not sure why people think AMD has an insurmountable advantage.
English
2
0
16
632
JustBenching
JustBenching@lackynamber1234·
This is the average across 14 games. Say we swap out the 2 extreme outliers (BG3 and AC) and replace them with 2 extreme outliers for Intel (TTW - beamng - cities skylines) it would not be unreasonable for Intel to have the gaming lead - especially in 1% lows. Wild.
JustBenching tweet media
English
15
0
32
4.3K
meng
meng@meng59739449·
18A NVL 8+16 (Standard Compute Tile) ~110mm² NVL 8+16+144 MB (bLLC Compute Tile)~150mm² NVL 16+32 (Dual Compute Tile)~220mm² NVL 16+32+288 MB (bLLC Dual Compute Tile) ~300mm² N4P AMD Zen 5 8 Core CCD + 32 MB/64 MB X3D~71mm² N2 AMD Zen 6 12 Core CCD + 48 MB/TBD MB X3DL3 = ~76mm²
Română
7
2
96
7.9K
Alex
Alex@Alex_Intel_·
I'm glad SemiAnalysis finally leaked 18A-U Now I can talk about it lol. Anyway 14A is coming earlier than 2029 guys but keep believing Lip-Bu sandbagging
Alex tweet media
English
8
14
133
11.7K
Tex
Tex@TEX_cs·
@kopite7kimi Are they clocking all 32 E-cores to 5.5ghz ? Wtf is that shit. I like it btw.
English
0
0
7
915
kopite7kimi
kopite7kimi@kopite7kimi·
The power consumption of a full-load NVL-K is over 700 watts.
English
52
28
206
49.1K
MCH
MCH@MCH2024·
@insane_analyst Intel fixed their chiplet latency issues with PantherLake. ARL/MTL had their CPU connected to the IMC by a paper straw. 2.5 GHz clock and no SLC. PTL fixes it. 5.5 GHz and 8MB SLC. Even D2D clock is up 50%. Result: -45ns memory latency. Big for NVL-S.
MCH tweet media
English
6
5
36
1.4K
Tex
Tex@TEX_cs·
@OneRaichu Any idea how much power LNL pulls for P-core fmax ?
English
1
0
0
191
Tex
Tex@TEX_cs·
@Bullsh1t_buster If true, then the chip size is small right ? So you should say "Small, if true".
English
0
0
1
134
High Yield
High Yield@highyieldYT·
Updated @NVIDIAGeForce RTX 50 Super line-up leaked 🚨 - RTX 5090 Super 16GB - RTX 5080 Super 8GB - RTX 5070 Super 6GB - RTX 5060 Super 4GB
English
173
161
5.4K
765.9K
Tex
Tex@TEX_cs·
@harukaze5719 ARL-R with the magic juice will cure all problems. /s
English
0
0
1
511
포시포시
포시포시@harukaze5719·
Everybody says great iGPU performance of B390. But what about the center of Intel, the CPU performance. It would be a bridge for the next gen, NVL So, what about the CPU Core's game performance? Don't tell about ARL-R lol
English
7
2
68
7.3K
Tex
Tex@TEX_cs·
@3DCenter_org @GohlJoao You cannot overclock it. So unless you are getting the highest end skus, you will have low boost clocks. Better to get the regular Raptor lake.
English
1
0
1
114
3DCenter.org
3DCenter.org@3DCenter_org·
@GohlJoao It *is* not, indeed. But there is no reason, why it *can* be not.
English
2
0
4
236
3DCenter.org
3DCenter.org@3DCenter_org·
Another possibility for a CPU upgrade for DDR4 owners in 2026: Bartlett Lake for desktop/consumer. The CPU exists, offers DDR4 support, and runs on LGA1700. No further technology is required; production just needs to be ramped up — and Intel needs to have the courage to do so.
English
3
3
34
3.1K
Tex
Tex@TEX_cs·
@meng59739449 Up to 20 and 30 ? I expect no more than 5-7% range for gaming perf improvements.
English
1
0
3
501
meng
meng@meng59739449·
Arrow Lake Refresh new 1)Compatibility with Arrow Lake motherboard 2)Ultra Plus performance up to 10+ or above 3)Some gaming performance up to 20 and 30 4)Launch in March 2026 5)Same price with Arrow Lake 6)Looks like redesign IO tile
English
3
1
37
5.8K
Tex
Tex@TEX_cs·
@Haze2K1 So 2 i9s and 2 i7s with bLLC.
English
0
0
1
871
Haze
Haze@Haze2K1·
Nova lake-s with big last level cache: 16p 32e 4lpe 14p 24e 4lpe 8p 16e 4lpe 8p 12e 4lpe
HT
6
6
49
8.3K
Tex
Tex@TEX_cs·
@Silicon_Fly Compared to LNL i9, I get ~5% better ST score. Considering the same ST boost of 5ghz, the improvement is entirely from ipc gain. So Cougar Cove is ~5% better than Lion Cove. Comparison is fair because both PTL and LNL have on-die imc.
English
0
0
4
281
SiliconFly
SiliconFly@Silicon_Fly·
Finally!!! Panther Lake Top SKU 388H performance revealed! And it's actually better than expected! Performance is actually a fair bit faster than ARL 285H in ST/MT. And freq is 5.1 GHz which is inline with previous leaks. This puts a serious damper on the entire Ryzen AI Max series (including Ryzen AI Max+ 395, Ryzen AI Max HX 375/370, etc). PTL is going to be the best x86 Laptop CPUs in 2026. Good performance, Extremely efficiency, Terrific Gaming, Fast AI & a lot cheaper than a MacBook. Intel finally has a clear winner. Thanks to 18A. Intel Laptop market share is now at ~78%. It should easily cross 80% next year.
SiliconFly tweet media
Benchleaks@BenchLeaks

[GB6 CPU] Unknown CPU CPU: Intel Core Ultra X9 388H (16C 16T) Min/Max/Avg: 4042/5077/5039 MHz CPUID: C06C2 (GenuineIntel) Single: 3057 Multi: 17687 browser.geekbench.com/v6/cpu/15474224

English
14
11
108
15K
Tex
Tex@TEX_cs·
@MCH2024 @Silicon_Fly I just checked 285H scores. In GB6 ST it scores ~2600. Which is lower than LNL 288V's score of ~2900. Despite higher clocks, ARL-H still shits the bed in GB6. PTL-H has the advantage of on-die imc and SA. So the score is better.
English
1
0
0
70
MCH
MCH@MCH2024·
@Silicon_Fly It’s similar to 285H actually in ST/MT. Slightly less but I am sure 388H still has a few hundred points to squeeze. The big thing is PTL beats STX Point pretty comfortably at MT.
English
4
0
6
572